Anthropic launched Fable 5 on June 9, 2026. It was the first time the company let the public touch its most capable tier, the Mythos class, the line it had previously called too dangerous in the cybersecurity domain to ship. Three days later, at 5:21 p.m. Eastern, a letter arrived from the Commerce Secretary. By that evening Fable 5 was gone. Not throttled, not geofenced. Gone, worldwide, for every customer. The less restricted sibling, Mythos 5, the version only a vetted set of partners could use, got pulled along with it. The stated trigger was a reported way to jailbreak the model, a trick that amounted to asking it to read a codebase and find the flaws.

Almost immediately, people read this as the smoking gun. The models have gotten so strong the government is yanking them off the shelf, the argument went, and behind the lab firewall those same systems are quietly rewriting themselves into something far past human. The shutdown was proof. The takeoff had started.

Follow that idea for a second, because it falls apart in your hands. The government did not lock Fable in a vault and let it cook. It cut its power and shoved it out of reach. The version the public never had, the one reserved for trusted partners, got killed in the same stroke. If frontier labs were sitting on a self-improving superintelligence, the Fable episode would be strange evidence for it, because the machine here was a kill switch, not a greenhouse.

That gap, between what the shutdown felt like and what it was, is the whole subject. The raw facts of 2026 are more dramatic than most people outside the labs realize. The story laid on top of them usually runs a step or two past where the facts can carry you. The interesting work is finding exactly where the facts stop and the story begins, because that line is where the money and the risk both live.

So let me take the question seriously and from both ends. Is predicting the next token enough to get us all the way to superintelligence, the kind that redesigns itself faster than we can watch? And are we, right now, inside the fast and terrifying version of that, the hard takeoff, or the slower one we can still steer?

What a next-token predictor actually does

Strip away the marketing and a large language model does one small thing, over and over. Given a string of text, it guesses the next chunk. Then it adds that chunk and guesses again. Training it means showing it a huge pile of human writing and nudging it, billions of times, to make that guess less wrong. That is the whole objective. Everything else, the essays, the code, the legal analysis, falls out of getting very good at that one guess.

The skeptics’ oldest line is that this can only ever be mimicry. A system trained to predict words is matching surface patterns in text, and patterns in text are not the world. A house cat understands gravity. It plans a jump, predicts where a falling object will land, models cause and effect, and it has read exactly zero words. A language model can write you a flawless paragraph about gravity and could not catch a ball. Yann LeCun, who ran AI at Meta until he left in late 2025 to build a company around a different design, has made this case as bluntly as anyone. He thinks autoregressive models, the technical name for these next-token machines, are a path that climbs impressively and then dead-ends short of the summit. His proposed replacement learns by predicting abstract states of the world rather than words, and his bet is funded to the tune of about a billion dollars, which is a useful reminder that the smartest skeptic in the room is not a crank.

Here is the catch that keeps the skeptics from closing the case. When researchers trained a small model only to predict the next legal move in the board game Othello, and then went looking inside it, they found something it was never told to build: a representation of the board. The model had no eyes and no rules. It saw only strings of moves. To predict the next move well, it had reconstructed the thing the moves were about. You can probe that internal board, flip a piece in the model’s “mind,” and watch its predictions change accordingly. That result, and a stack of others like it, is why the cleanest version of the mimicry argument fails. To predict well enough, the system is pressured to understand. Ilya Sutskever, who has as much claim as anyone to having seen this from the inside, has made the same point in plain terms: predicting the next token well means understanding the reality that produced it.

That is the deep reason the next-token bet is not obviously stupid. Compression is comprehension. If you can predict something, you have modeled it. My confidence that these systems build real internal models of the things they discuss is high. My confidence that text-only prediction builds a model rich enough for full general intelligence is much lower, and that is the crack LeCun keeps his thumb in.

The thing the skeptics got wrong, and the thing they got right

For a few years the recipe was simple. Make the model bigger, feed it more text, give it more computers, and it got better on a smooth and almost embarrassingly predictable curve. People called these the scaling laws, and they held across several jumps in size. That era is closing.

OpenAI’s big 2025 pretraining run, the one that shipped as GPT-4.5, was the tell. It cost an enormous amount and delivered a much smaller jump than the prior generation had. Sutskever said the quiet thing at a December 2024 talk: pretraining as we have known it will end, because data is the fuel and we have only one internet. The supply of high-quality human text is finite, and the largest runs are already drinking from the bottom of the glass. The skeptics who said “the scaling wall is real” were right.

What they missed is where the road turned. The labs stopped trying to win only by making the model bigger and started spending more at the moment of use. Instead of answering instantly, the newest models think first. They write out a long internal chain of steps, check their own work, notice a wrong turn, back up, and try again before they answer. OpenAI’s o-series and DeepSeek’s R1 are built this way, trained by reinforcement learning, which means the model gets rewarded for reasoning that reaches correct answers and learns to do more of it. This is still a next-token predictor. It is the same engine, now allowed to talk to itself on the way to a reply, and rewarded for talking itself into better answers.

Notice what that resembles. When a model writes out its reasoning, evaluates it, catches its own error, and corrects course, it is doing a crude version of the loop people point to when they say humans are more than pattern matchers. We deliberate. We hold a thought, inspect it, argue with it, and revise. The reasoning models externalize that loop onto the page in tokens. It is the difference between a streetballer who reacts and a point guard who reads the defense, runs two options in his head, and picks the better one before he moves. Whether the model’s version is genuine reasoning or fast retrieval dressed as reasoning is a real fight, and the honest answer is that it is some of both, in proportions nobody can yet measure. Moderate confidence that the reasoning is partly real, and low confidence on how far it generalizes past the patterns it was trained on.

This matters for the central question because it changes what “all you need” means. If you had asked in 2024 whether scaling the next-token predictor was enough, the honest answer was becoming no. The pretraining curve was bending. But the paradigm did not die. It grew a second engine, the thinking-at-inference engine, and that engine is young and its own curve has not bent yet. So the question is not settled by the pretraining slowdown the way the skeptics hoped. The bet just moved to a different table.

The strongest card the bulls hold

Here is where the facts get genuinely hard to wave away, and where I had to verify every number before I’d repeat it, because the loose versions floating around are inflated.

In May 2026, Anthropic published its own internal data on how much of its work the AI now does. More than 80% of the code merged into Anthropic’s own codebase was written by Claude. Before its coding agent launched in early 2025, that figure sat in the low single digits. The company’s leadership puts the looser number, counting scripts and throwaway code, north of 90%. The output per engineer tells the same story: in the second quarter of 2026 a typical Anthropic engineer was shipping roughly eight times the code per day they shipped in 2024. One engineer, the company reports, had not written a line by hand in five months.

The capability behind that is climbing on a curve worth staring at. The length of task a model can finish on its own, with no human stepping in, has been doubling about every four months, up from every seven. In March 2024 the best model could handle a software task that takes a person about four minutes. A year later, about ninety minutes. By 2026, twelve-hour tasks. A research preview, the unreleased Mythos model, ran for at least sixteen hours on its own, which is the edge of what the outside evaluators could even measure. On one optimization problem, that preview found a 52x speedup where a strong human researcher typically gets about 4x in a half-day of work.

And the AI is starting to improve the AI. Google’s AlphaEvolve, a system that uses models to search for better algorithms, found a way to multiply a certain class of matrices using 48 multiplications instead of 49. That sounds trivial until you learn the previous record stood since 1969, for 56 years, and that the result is mathematically verifiable, not a vibe. The same system claws back about 0.7% of Google’s entire worldwide computing fleet by scheduling it better, and sped up a key training routine enough to cut roughly 1% off the time to train Google’s flagship model. A model, helping train its successor. That is the loop everyone is watching for, observed in a small but real form.

If you wanted to make the case that we are in or near a hard takeoff, this is the case. The thing improves. The thing now helps build the next thing. The task horizon is doubling fast. The forecasters who called this are not all cranks either. Leopold Aschenbrenner, who wrote a widely read 2024 essay arguing the trendlines pointed to roughly human-level AI by 2027 and then a fast intelligence explosion, got the infrastructure story very right; the trillion-dollar buildout he predicted is happening, and he now runs a hedge fund betting on the picks-and-shovels of it. Ray Kurzweil has held to AI matching humans by 2029 and a full singularity around 2045, and his 2029 date, once fringe, now sits inside the range serious lab leaders give. The “AI 2027” scenario laid out a month-by-month path through an automated-research explosion that, read in 2026, does not feel like science fiction.

The thing the bulls keep getting wrong

Now turn the same facts over and look at the underside.

Start with that 80% figure, because it is the one people quote most and understand least. It is Anthropic’s number for Anthropic’s codebase, not an industry truth, and writing code is the single task these models are best at, the home court. Software has a property most work lacks: you can check the answer automatically. The code runs or it doesn’t, the test passes or it doesn’t, and that clean signal is exactly what reinforcement learning needs to train on. The horizon doubling every four months is measured mostly on coding and technical tasks for the same reason. None of this tells you the model is as far along at the messy, unverifiable work that fills most of the economy, the negotiation, the judgment call with no test suite, the decision about what is even worth doing.

That last one is the wall the labs keep hitting, and to their credit they say so. Anthropic’s own paper, the one with the 80% number, is built around the idea of AI building itself, and its own researchers write that recursive self-improvement “is not here, nor is it inevitable.” Their most likely scenario is not the runaway. It is a world where the AI does more and more of the doing while humans keep setting the direction. The bottleneck they name is research taste, the senior-level judgment about which problem to chase and which result to trust. In late 2025 the model picked a better next research step than the human about half the time; months later, closer to two-thirds. Climbing, clearly. Closed, no.

It is worth being skeptical even of the impressive anecdotes. The paper’s showcase example, a model that diagnosed a nasty production incident in two hours that would have taken a person days, is a real and useful thing. It is also, when you read it closely, classic debugging: a clear problem, rich error data, a fix to be found. A model finding an obscure flag faster than a tired human is the compiler catching your typo, scaled up. It is enormously valuable. It is not the same as the model deciding, unprompted, what the company should build next quarter, which is the capability that would actually close the loop.

Then there is generalization, the soft spot under all of it. There is a test called ARC-AGI, built specifically to be easy for humans and hard for memorized knowledge. Its second version, released in 2025, knocked frontier models down to near zero at launch while ordinary people solved the puzzles without much trouble. Scores have since climbed, at high cost, which tells you the wall is scalable but not free. Apple put out a paper showing that reasoning models, pushed past a certain complexity, do not degrade gracefully; they collapse, and stranger still, they sometimes try less as the problem gets harder. The rebuttals were fair, some of Apple’s puzzles were rigged in ways that guaranteed failure, but the core point survived the fight. These systems have a frontier of difficulty past which the reasoning stops being reasoning, and we do not know how to push that frontier reliably with scale alone. Moderate-to-high confidence that this limit is real; moderate confidence on how binding it stays as the inference-time engine matures.

Now back to the opening, because the causal story behind the hard-takeoff thesis is where it breaks hardest. The claim is that government is pulling the best models off the market, which lets the labs keep those models internal, which means recursive self-improvement is happening behind the firewall, out of view. The June 2 executive order is voluntary; it explicitly does not create a licensing or pre-clearance requirement, and the 30-day window it describes gives the government an early look, not the lab a private runway. The Fable shutdown removed the model from everyone, including the labs’ own foreign-national staff. When the government restricted OpenAI’s GPT-5.6 later that month to about twenty approved partners, OpenAI pushed back in public, saying this kind of gating “should not be the long-term default.” The labs are fighting to release these models, not hoarding them to self-improve in the dark. The internal-versus-public capability gap is real, and it is mostly explained by dull things: safety testing, the cost of serving a model to hundreds of millions of people, and the obvious competitive logic of not handing rivals your sharpest research accelerant. A gap that boring is not evidence of a secret intelligence explosion. It is evidence of caution and economics.

And the physical world is slower than the digital one by a margin that bounds how fast any of this can hit the economy. Roughly half the AI data centers planned for 2026 in the United States have slipped or been canceled. Transformers are back-ordered, the power grid is strained, and only about a third of the new capacity people projected is actually under construction. You can have an algorithmic breakthrough on a Tuesday. You cannot conjure a gigawatt of power and the steel to use it on a Wednesday. Even Aschenbrenner’s own thesis leans on this; his hedge-fund bet is less about clever code and more about electrons, because electrons are the constraint that clever code runs into.

The forecasters are slipping accordingly. The “AI 2027” authors have quietly walked their median toward 2029 and 2030. Aschenbrenner’s revenue prediction for mid-2026 came in well under his line, and his call that open-source models would fade was flatly wrong; the cheap open models from China are sitting right behind the frontier and forcing prices down. Being early and being wrong are different, and these are early. But early enough that anyone underwriting against 2027 as a date should stop.

So which is it

Both pictures are true, which is why smart people keep talking past each other. The capability is compounding faster than the skeptics admit. The runaway is further off than the bulls claim. The shape that fits the evidence is a fast soft takeoff: a steep, accelerating ramp where the AI gets dramatically more capable and starts meaningfully speeding up its own development, while humans stay in the loop at the points that matter and the physical world throttles how fast any of it reaches the ground. The loop is real and it is open. Closing it requires automating the senior judgment that the labs themselves admit they have not automated, and clearing a generalization wall we do not know how to clear on command.

My honest probabilities, held loosely: fast soft takeoff as the base case, the most likely world by a comfortable margin. A genuine hard takeoff, the weeks-to-months runaway, as a real tail, somewhere in the rough range of one in seven to one in four this decade, mostly through the automated-research channel if that research-taste gap closes faster than the physical constraints bite. Not negligible. Not the base case. Anyone who tells you they know which of these we are in with confidence is selling something, possibly a fund.

On the title question, then. Is the next token all you need? For an AI that is superhuman across most of what can be checked and verified, probably yes, and we are most of the way there. For full self-improving superintelligence, unproven, and the honest word is unproven rather than no, because the next-token engine keeps doing things its critics swore it couldn’t, and because the inference-time reasoning engine bolted onto it is too young to have shown its ceiling.

One more thing, on the seductive line that humans are just next-token predictors too. There is real science under it. A leading theory in neuroscience holds that the brain is fundamentally a prediction machine, constantly guessing its next sensory input and learning from the error. When you read this sentence, your brain is predicting the word before your eyes reach it. The rhyme with a language model is not an accident. But the brain predicts a flood of sound and sight and touch and its own movement, grounded in a body, on about twenty watts, learning continuously as it goes. It was not trained by gradient descent on the internet. Calling a human a next-token predictor is a metaphor that illuminates one shared trick and hides a dozen differences that might be the whole game. Use it to understand why the bet is plausible. Do not mistake it for a proof that the bet pays.

What this means if you allocate capital

Switch tracks now, from what is true to what to do about it, because the two get jumbled and both suffer for it.

The money has already voted, hard. In the first quarter of 2026, global venture funding hit about $300 billion, and roughly 80% of it, around $240 billion, went to AI. Four rounds, OpenAI at $122 billion, Anthropic at $30 billion, xAI at $20 billion, and Waymo at $16 billion, took nearly two-thirds of all venture dollars on Earth. A single quarter of AI funding exceeded all of 2025. That is not a sector. That is a gravity well.

Here is the part most people get backwards, and it is the same backwards as the Fable story. A soft takeoff is the harder world to invest in, not the easier one. If a hard takeoff were imminent, almost nothing you funded at the application layer would matter, because a superintelligence would eat every workflow at once and the only sane bets would be the compute and the power underneath it. The soft takeoff is more demanding precisely because the technology keeps getting better and cheaper underneath every company you back, on a schedule, for years. The model that makes your portfolio company magic this year is the commodity that makes it ordinary next year. The cost of a million tokens fell about 80% from 2023 to 2025. Any business whose only edge was reselling access to a model has already watched its margin evaporate.

So the test experienced investors are starting to apply is brutal and simple. Two questions, and a company needs a real answer to both:

• If a frontier lab shipped your exact product as a default feature in their next release, would your customers cancel?

• Does what you do survive three more model generations getting cheaper and smarter?

Most thin wrappers around someone else’s model fail both. What passes tends to own something the model maker cannot reach from a data center. Proprietary data that compounds the more the product gets used. A workflow so embedded in how a company runs that ripping it out costs more than tolerating it, which is why coding tools, legal-research tools, and enterprise-search tools that became the system of record have held up while generic chat skins have not. Trust and compliance in regulated corners, healthcare, finance, defense, where being right and being accountable matter more than being clever. And the physical world, robots and the machinery of moving atoms, where the bottleneck is not tokens and the frontier labs have no special advantage.

The two layers worth real money look opposite and are both defensible. One end is the picks and shovels: compute, power, memory, the boring infrastructure the whole boom runs on, which pays off in almost every scenario including the scary one. The other end is the deep vertical application that owns its data and its customer so completely that a better base model helps it rather than kills it. The undifferentiated middle, the layer that is just a prompt and a logo on top of an API, is where capital goes to die, and it is where a frightening amount of 2026 seed money is going anyway.

For the people who fund the funds, the limited partners, the uncomfortable truth is that AI is your largest exposure whether you chose it or not, through the venture portfolios, through the public indices where the ten biggest companies now make up a share of the market that exceeds the peak before the 2000 crash. The question is not whether to be exposed. It is whether the exposure is concentrated in things that survive a soft takeoff, where defensibility and timing decide everything, or scattered across things that the next model release quietly deletes. The barbell, infrastructure on one end and defensible verticals on the other, with as little as possible in the middle, is the shape that respects both the upside and the wall.

And the macro claim you will hear at the top of every keynote, that AI and robots will multiply the global economy tenfold in a decade, the Musk number: low confidence, and I would bet against it on that timeline. The serious range from people who model growth for a living runs from a rounding error to something genuinely large, several percentage points of added output over a decade in the credible middle, with explosive growth as a real but later and far less certain tail. A tenfold jump in ten years requires the hard-takeoff world and the physical buildout to both arrive on schedule, and the physical buildout is already slipping. Plan for a serious productivity boom. Do not underwrite a miracle.

The shutdown on that Friday in June is the whole thing in miniature. A model good enough to scare a government (encouraged by fear based marketing by the labs), killed by a letter, pushed out of reach instead of locked away to grow. Powerful and constrained at the same time, the capability sprinting while the institutions and the power grid jog to keep up. That is the texture of a soft takeoff. It is less cinematic than the runaway, and harder to live inside, because it asks you to keep making good decisions year after year while the ground moves under you, instead of betting everything once on a single discontinuity. The next token might well be most of what you need. It is not, yet, all of what it would take to stop needing us. The interesting years are the ones in between, and we are in them.

An agent is a program that uses an AI model to do work on its own. It queries, retries, checks itself, and loops, the way a tireless junior analyst might if the analyst never slept and never got bored. Point a fleet of them at a conversational data tool like the one Databricks sells and they hammer the underlying database around the clock. Every query costs something to serve. Multiply by millions of queries and a six-point hole opens in the margin.

Sit with that, because it flips the oldest fact in the software business. The reason software became the best business model ever invented is that once you have built the thing, serving one more user costs almost nothing. Usage was free money. A customer who used your product twice as much made you richer at no extra cost. That is the entire reason a good software company can earn 80 cents on the dollar.

Agents changed the sign on that equation. Now the heaviest users are the ones quietly draining you, because behind every agent action sits a model call, and model calls cost real money every single time. The unit of consumption is the token, which is roughly a syllable of text the model reads or writes, and you pay per token whether the work was useful or the agent just spun in a circle. The meter is always running now. This is the thing the whole week was secretly about.

A budget is the first thing to break

You can watch the same shift hit from the buyer’s side. Uber, a company that instruments everything and is not in the habit of losing track of a cost, reportedly gave its roughly five thousand engineers a monthly allowance for AI coding tools, somewhere between $500 to $2,000 dollars a head. By April they had spent the entire year’s budget. The company then capped it. Walmart, Cisco, Amazon, and Meta reportedly did versions of the same thing.

When a company that good at counting misjudges a number by a factor of three, the number is new. Nobody overruns the electricity bill by 3x, because everyone has been paying electricity bills for a hundred years and the intuitions are baked in. AI spend has no baked-in intuition yet. It was modeled like a software seat, a fixed price per person per month, and it behaved like an appetite. Give a smart engineer an agent that can run a hundred experiments overnight and the engineer will run a hundred experiments overnight, because why wouldn’t they. The tool got good, so people used it more, so it cost more. The better it works, the more it costs, which is a sentence that has never been true about software until right now.

Databricks’ own response tells you how serious this is. The company built an internal reinforcement-learning tool, reportedly called KARL, whose entire job is to make its customers’ agents query less expensively. When a software vendor has to ship a feature to stop its own product from being used too hard, the economics have moved under everyone’s feet.

Rationing from the top

Here is where the week turns from a finance story into a power story. The same week CFOs started rationing tokens from the demand side, the government started rationing models from the supply side.

OpenAI’s newest model, the GPT-5.6 line, did not get a normal launch. It went out under a review where the federal government cleared access customer by customer, with Commerce sitting in the middle deciding who was allowed in. Anthropic’s most capable models were restricted to a small set of vetted cybersecurity and infrastructure firms under a defensive program, after an earlier outright block. For the first time, the best American models are shipping by permission rather than by press release.

The reason is not hard to see. These models are now good enough at finding software vulnerabilities and writing exploits that handing the strongest version to anyone with a credit card looks, from a national-security desk, like handing out a capability rather than a product. Whether that framing is right is a separate argument (although this is exactly what Anthropic’s fear based marketing was asking for). What matters for anyone building a company is that frontier-model access just became a thing a government can switch off on a Tuesday.

Alex Wissner-Gross, the researcher who writes The Innermost Loop, made the sharp observation about what gating actually does. Slowing how fast a lab is allowed to ship does nothing to slow how fast it is allowed to train. So the distance between what the public can use and what exists privately inside the labs does not hold steady. It widens. Every week the public frontier is held back, the real frontier keeps moving, and the gap becomes a strategic asset owned by a handful of organizations and their chosen customers.

Put the three pieces side by side and they rhyme. A CFO rationing tokens by budget. A vendor rationing queries with a throttling tool. A government rationing the model itself by clearance. Three different actors, same underlying move, because the same thing happened to all of them. A resource that felt free when it was a demo became scarce the moment it became a workload that runs continuously. And scarce things get rationed by whoever controls the choke point.

Who wins when the meter rules

If usage is the cost, then the winner is whoever can afford to let the meter run without flinching. That is a very different winner than the one most people were betting on a year ago.

The clearest example is the strangest one. SpaceX, fresh off going public, used its newly liquid stock to buy Cursor (as planned), the AI coding editor, in an all-stock deal reported at around sixty billion dollars, after having already absorbed xAI. A rocket company now owns the satellite network, the supercomputer, the model, and the screen the developer types into. People keep asking why a space company is winning the coding-tool war. The answer is the meter. SpaceX can let agents run flat out because it owns the electricity and the silicon underneath them. It is not watching the token bill the way a venture-funded wrapper has to. When the cost of a product is compute and power, the company that owns compute and power can simply outlast everyone selling a thin layer on top.

Notice the deal was paid in stock, not cash. That is its own signal. When the strongest player in the market pays for a sixty-billion-dollar acquisition with its own shares instead of money, it is telling you what it thinks those shares are worth, and it is conserving the cash for the thing that actually constrains it, which is building more compute and securing more power. (Worth noting the acquisition valuation was secured at a lower SpaceX valuation so effectively they paid something like half the advertised priced!)

The talent flows confirm the same gravity. John Jumper, who shared a Nobel Prize for AlphaFold, left Google DeepMind. Noam Shazeer, a co-author of the original transformer paper and a researcher Google reportedly paid billions to bring back in 2024, left for OpenAI. Within days of each other, two of the most valuable researchers alive walked out of a diversified giant and into pure-play labs. The best people are choosing the places that are allowed to push the frontier hardest, which widens the capability gap, which makes the pure-play labs even more attractive to the next departure. The loop feeds itself.

The constraint that everyone underweights is power

Follow the meter far enough and it stops being about money and starts being about electricity. Agents that run continuously are workloads that run continuously, and workloads need power. Projections for data-center electricity demand keep getting revised up, fast enough that the binding limit on the next phase of AI may turn out to be the grid rather than the model.

This is why the orbital-compute conversation stopped sounding like science fiction this week. On the Moonshots podcast, Planet Labs’ Will Marshall and others walked through the case for running AI inference on satellites, processing remote-sensing data in space and only beaming down the conclusions, paired with the idea of powering data centers with continuous solar arrays in orbit where the sun never sets. You can dismiss the specifics. The underlying instinct is correct. When power becomes the constraint, people start looking for power in places they previously ignored, including up. The serious money in AI is migrating toward whoever can generate, secure, and cool electricity at scale, and that is a very physical, very capital-heavy place for a software boom to end up.

What this means for founders

Separate the building advice from the investing advice, because they point in slightly different directions.

If you are building, the meter reorganizes what is worth building. The wedges that got more attractive this week are the ones that either dodge the meter, control it, or survive someone else flipping a switch.

A routing and orchestration layer that lets an application swap between a US model and an open-weight one in real time is now close to mandatory infrastructure, because the alternative is having your product bricked the day your single provider gets restricted. On-device and edge inference got more interesting for the same reason, since running the model locally sidesteps both the cloud bill and the question of who is legally allowed to call the API. Anything that throttles, caches, or optimizes agent spend is selling directly into the wound Databricks just showed everyone, so cost-control tooling went from a nice-to-have to a budget line. And dual-use work tied to compute, secure communications, and energy lines up with where the capital and the procurement dollars are actually flowing.

The wedges that got less attractive are the mirror image. A single-model wrapper with no proprietary data is now squeezed from above by the platforms bundling the same feature and from the side by the model getting restricted out from under it. Flat-priced agentic software sitting on top of uncapped API spend is a margin trap waiting to spring, because your heaviest users are your biggest losses. And services built on manual offshore QA or junior-developer outsourcing are racing automated patching that gets cheaper every month.

The questions worth forcing in a pitch this week are concrete:

• If your model provider gets restricted on Tuesday, does your product still work on Wednesday?

• Does your pricing survive a customer whose agent runs a thousand times more often than a human would?

• Are your heaviest users your most profitable accounts or your biggest losses, and can you tell me the number?

• Can a foreign-national engineer on your team legally touch your primary model API next quarter, and what breaks if the answer becomes no?

• Are you selling a capability demo, or a line item a CFO will defend during the next budget cut?

Have you tried the new open-weight models?

There is a question worth running before you raise another dollar to cover your model bill: have you actually tested the open-weight models lately, or are you paying frontier prices out of habit?

The gap between the best closed model and the best open one has been collapsing on exactly the tasks most startups run, the coding, extraction, summarization, and routing work that makes up the bulk of real production traffic. The frontier still wins at the hard edge, the long-horizon reasoning and the genuinely novel problem. Most products do not live at that edge. They live in the middle, doing the same bounded task ten million times, and in the middle an open model running on rented hardware can deliver something close to the same answer for a fraction of the price. The arithmetic that matters is blunt: if a capable open-weight model gets you 95 percent of the value at 85 percent lower cost, the 5 percent you are paying frontier rates to recover has to be worth more than the margin you are burning to get it. For most workloads it is not.

So measure it instead of assuming it. Take your actual production traffic, not a benchmark, and replay a representative slice through an open model. Score the outputs against what your frontier provider returns. You will usually find the work splits cleanly. A large share comes back indistinguishable, a slice is good enough with light guardrails, and a thin tail genuinely needs the frontier. Route on that finding. Send the bulk to the cheap model, keep the frontier for the tail, and you have just rebuilt your cost structure without touching the product the customer sees.

The economics compound past the per-token savings. An open model you can host or fine-tune is one nobody can restrict out from under you on a Tuesday, which is no longer a hypothetical given how this month went. It is one you can run at the edge or on-prem for customers who care where their data sits. And it is one whose cost you control rather than rent, which means your margin stops being a decision your model provider makes for you. The companies that win the next year will not be the ones using the smartest model. They will be the ones who figured out the cheapest model that clears the bar for each job, and who built the routing to put every request in front of the right one.

The reflex to default to the frontier for everything was rational when the gap was wide and the price difference was small. Both halves of that flipped. If you have not re-run the test in the last quarter, your cost structure is built on a price comparison that is already stale.

What this means for LPs

The week sharpened a dispersion that has been building for months. Capital is concentrating violently into a few platforms that own compute and power, while the messy middle of enterprise software gets repriced down toward acquisition value. That gap is the whole environment, and it argues for two disciplines at once.

At the early stage, the discipline is to fund the throttle and the portability layer and to refuse the high-burn wrapper, because the wrapper is exactly the business model the meter punishes. A fund built to write small seed checks into model-portable, cost-aware, infrastructure-adjacent companies is positioned for this. A fund chasing flashy consumer agents with uncapped token bills is underwriting margin collapse and has not noticed yet.

In the secondary book, the discipline is to price on real numbers rather than momentum. Databricks growing past a multi-billion-dollar revenue run rate while its margin compresses to 74 percent is the tell. Even the winners are absorbing the same cost shock, so the right entry mark is a function of revenue quality and gross margin, not the most recent post-IPO headline. Funding is a price signal, never a quality signal. The SpaceX print is dazzling, and the correct response to a dazzling print is discipline, not fear of missing out.

What this means for the venture market

Three structural facts surfaced this week that change how the asset class behaves.

Liquidity came back, and it came back concentrated. The exits and the up-rounds are clustering in a handful of names while the long tail waits. Access to those names matters less than the price you pay to get in.

Acquisitions are being paid in inflated equity. A sixty-billion-dollar all-stock deal is a barter transaction between two richly valued private currencies, and it tells you the acquirer would rather spend paper than cash. Read those marks as directional, not precise.

The IPO line is real but orderly. The strongest names appear to be staggering their public debuts so they do not flood the same investor base at once, which means the public-market repricing of private tech will arrive in waves rather than a single reckoning. That gives a disciplined secondary investor time, and time is the one thing a momentum chaser never uses well.

What this means for VCs

The meter changes diligence before it changes anything else. For two years the central question in an AI deal was whether the product worked. That question is close to free now, because the models got good enough that most demos work. The question that separates winners from margin traps is whether the company makes money when the product works hard. So the diligence that matters moved from the demo to the bill. Ask for token cost per unit of delivered work, ask how it trends as usage scales, and treat any founder who cannot answer in those terms the way you would treat a SaaS founder in 2015 who could not tell you their gross margin.

The harder truth is that a lot of what looked like product moat this year was the model doing the work, and the model is rented. If a startup’s edge is capability the foundation lab can ship in its next release or a regulator can switch off by name, you are underwriting a lease, not an asset. The durable edge sits in the places the meter creates: proprietary data the model cannot get elsewhere, a workflow lock that survives a model swap, distribution into a budget line a buyer will defend, and unit economics that improve rather than decay as the agents run. Price the lease cheaply. Pay up only for the asset.

Check construction has to respect the new cost curve too. An early AI company now has a cost of goods that scales with adoption, which means a seed round that would have lasted eighteen months on a 2019 burn profile can evaporate in nine if the product takes off, because success spends compute. Founders who win burn cash by being used. Reserve accordingly, and stress the model against the good case, not just the bad one, since the good case is where the token bill explodes.

The questions worth forcing in a partner meeting this week:

• Does this company’s edge survive the next frontier-model release, or is it renting capability that the platform will absorb?

• What is the gross margin at scale once agents, not humans, are the primary users, and is the founder even measuring it?

• If the primary model provider gets restricted, does the company have a portability story or a dependency it has been calling a feature?

• Are we paying for an asset the company owns or a lease on someone else’s model?

On strategy, the consolidation cuts against the late-stage growth game and toward the early one. When a handful of platforms own compute, power, and distribution, the value they create accrues to them, and writing a growth check into that gravity well is buying a crowded, richly priced position. The unowned ground is early, in the throttle-and-portability layer the platforms have no incentive to build and the application wedges too small for them to chase. That is where a disciplined seed fund still gets real ownership at a price that leaves room. Funding remains a price signal and not a quality signal, and this week the loudest prices were in exactly the places where the next dollar is least likely to compound.

The scarcity moved

The assumption underneath the last two years was that if intelligence got cheap, it would get abundant, the way bandwidth did. Cheap bandwidth gave us streaming video and nobody thinks about the cost of a megabyte anymore. We expected the same arc for tokens.

It is going the other way. Intelligence is getting cheaper per token and more expensive in total, because we have learned to consume so much more of it (a Jevon’s paradox!), and the things it actually rests on are not getting cheaper at all. Power is getting scarcer. High-end silicon is getting scarcer. And permission, the right to run the very best models, just became something a government rations by name.

The scarcity did not disappear when the models got good. It moved. It used to live inside the model, in the cleverness that was hard to build. Now it lives in the meter, the grid, and the gate. The companies worth backing are the ones building where the scarcity actually went.

We already published a separate batch takeaways post with our broader impressions of the cohort, category-level observations, and what we think this batch says about where early-stage AI and software are heading. You can read that here:

Ignite Insights

What YC Spring 2026 Felt Like From the Room

Today is demo day…

Read more

15 days ago · 7 likes · Ignite Insights

This post is more direct: here are the companies we backed, what they do, and why we were excited to invest.

A few patterns stood out.

First, AI is moving from “copilot” to operating system. Many of the best companies are not building thin wrappers. They are trying to own full workflows: customer success, medical practice operations, construction estimating, pathology reporting, observability, marketing execution, sales development, and more.

Second, infrastructure remains a major theme. GPU utilization, HPC backtesting, app observability, and AI-native workflow systems are becoming more valuable as AI increases the volume and complexity of work.

Third, the batch was broader than generic SaaS. We invested across defense, aerospace, healthcare, fintech infrastructure, devtools, vertical AI, and applied AI agents. That matters because the biggest outcomes rarely look obvious at the start.

Our job is not to wait until consensus forms. Our job is to identify early signal, move quickly, and earn allocation before the round is gone.

The Companies We Backed (Alphabetically)

Arlo Industries

What they do: Arlo Industries is building a passive aerial sensing mesh to track drones, missiles, and other aerial threats without traditional radar.
YC Profile: https://www.ycombinator.com/companies/arlo-industries

Why we invested: Defense is being reshaped in real time by low-cost drones, autonomous systems, and asymmetric warfare. Legacy radar systems were not designed for a world where cheap aerial threats can appear everywhere at once. Arlo’s distributed, passive sensing architecture is compelling because it attacks both the technical problem and the economic problem: how do you create persistent, wide-area coverage without relying on centralized, expensive infrastructure? We were attracted to the severity of the problem, the timing, and the potential for Arlo to become a foundational sensing layer for modern defense.

Dispatch

What they do: Dispatch is building refurbishable reentry vehicles that can host and return payloads for companies manufacturing ultra-high-value materials in microgravity.
YC Profile: https://www.ycombinator.com/companies/dispatch

Why we invested: Space manufacturing is one of those categories that sounds futuristic until the enabling infrastructure arrives. The core bottleneck is simple: if companies can manufacture valuable materials in space, they still need a reliable way to bring those materials back to Earth. Dispatch is attacking that bottleneck directly. We liked the ambition, the technical depth, and the founder-market fit from a team with relevant spacecraft experience. This is a high-risk, capital-intensive category, but the upside case is enormous if Dispatch becomes a core logistics layer for in-space manufacturing.

Expanse

What they do: Expanse helps teams recover wasted GPU and HPC capacity by predicting the resources compute jobs actually need before they run.
YC Profile: https://www.ycombinator.com/companies/expanse

Why we invested: AI infrastructure spend is exploding, but much of that compute is still wasted through over-provisioning, failed jobs, and poor resource prediction. Expanse is building an intelligence layer for GPU and HPC infrastructure: read the job, understand the cluster, predict what resources are needed, and reduce waste before money is burned. The team has unusually strong founder-market fit, having worked directly on the kinds of HPC and GPU workloads they now serve. We invested because compute efficiency is becoming a strategic budget issue, and Expanse has a credible path to becoming a control point for cluster utilization.

InstaAgent

What they do: InstaAgent helps consumer brands create, distribute, test, and learn from large volumes of personalized social creative across personas and channels.
YC Profile: https://www.ycombinator.com/companies/instaagent

Why we invested: AI has made content creation cheap. It has not made content effective. That distinction matters. Consumer brands do not need more generic AI slop; they need high-quality creative variation, fast testing, and a learning loop that compounds across channels. InstaAgent’s wedge is scaled social creative generation and distribution, but the bigger opportunity is becoming a workflow layer for modern brand marketing. We liked the team’s speed, their understanding of social distribution, and the possibility that creative testing becomes more software-like as AI changes how brands operate.

Keyframe Labs

What they do: Keyframe Labs is building visual AI avatars and multimodal agents designed to become more scalable, expressive interfaces for AI applications.

YC Profile: https://www.ycombinator.com/companies/keyframe-labs

Why we invested: Voice AI was one major interface shift. Visual, embodied, multimodal AI may be the next. Keyframe is attacking the cost and scalability constraints that have historically made realistic avatars difficult to deploy widely. We liked the team’s technical ambition and the possibility that avatars become a core interface layer for sales, education, support, entertainment, and agentic software. If AI agents are going to represent companies, teach users, sell products, or guide workflows, they may need faces, presence, and visual interaction—not just text boxes and voice streams.

Klarify

What they do: Klarify is building AI software for therapists and mental health practices, starting with documentation and expanding into broader administrative workflows.
YC Profile: https://www.ycombinator.com/companies/klarify

Why we invested: The best vertical AI companies do not replace the professional; they remove the non-core work that prevents the professional from doing their highest-value job. Klarify’s insight is exactly that. Therapists are overburdened by notes, treatment plans, claims support, scheduling, payments, and client follow-up. Klarify keeps the human therapist at the center and uses AI to automate the surrounding operating burden. We liked the clarity of the wedge, the severity of the administrative pain, and the potential to expand from notes into the operating system for small and mid-sized mental health practices.

Memory Store

What they do: Memory Store is building memory infrastructure for AI applications, helping AI systems retain, retrieve, and use context over time.
YC Profile: https://www.ycombinator.com/companies/memory-store

Why we invested: Memory is one of the most important unsolved primitives in AI applications. Today, many AI products feel impressive in isolated interactions but weak across time because they do not remember enough, structure context well enough, or retrieve the right history at the right moment. Memory Store sits at a foundational layer: persistent memory for AI-native software. We invested because every serious AI workflow eventually needs durable context, personalization, and recall. If they become the memory layer for a meaningful share of agentic applications, the opportunity is large.

Oddpool

What they do: Oddpool is building data and infrastructure for prediction markets and event-based financial markets.
YC Profile: https://www.ycombinator.com/companies/oddpool

Why we invested: Prediction markets are moving from niche curiosity to serious financial infrastructure. As venues, assets, and trading strategies proliferate, institutions need normalized data, symbology, historical records, settlement metadata, and APIs they can build on. Oddpool’s opportunity is to become the neutral data and reference layer for this emerging market structure. We liked the infrastructure angle: as the category grows, high-quality historical data and normalized cross-venue infrastructure become harder to replicate and more valuable over time.

PerfectBit

What they do: PerfectBit is building AI infrastructure for software engineering and code intelligence.
YC Profile: https://www.ycombinator.com/companies/perfectbit

Why we invested: Software engineering is one of the first major labor markets being reshaped by AI, but the tooling stack is still early. The opportunity is not merely autocomplete. The larger prize is understanding codebases, automating complex engineering work, and helping technical teams ship faster with fewer bottlenecks. We liked PerfectBit because the category is enormous, the timing is right, and even small improvements in engineering productivity can create significant customer value. The risk is competition, but the market is so large that multiple durable companies can emerge.

PLAN0 AI

What they do: PLAN0 AI builds AI-native construction cost estimation and analytics software.
YC Profile: https://www.ycombinator.com/companies/plan0-ai

Why we invested: Construction cost estimation is slow, manual, expensive, and error-prone. It also sits at a critical point in the construction value chain: decisions made early can determine whether a project works economically years later. PLAN0 ingests floor plans and elevations, reconstructs projects, and helps produce cost estimates and scenario analysis far faster than traditional workflows. We invested because this is a severe, high-friction vertical problem with a clear AI wedge and a large potential expansion path into analytics, forecasting, and workflow ownership for construction teams.

Plena Health

What they do: Plena Health is building a full-stack operating system for specialty medical practices.
YC Profile: https://www.ycombinator.com/companies/plena-health

Why we invested: Specialty medical practices are operationally complex, understaffed, and buried under repetitive workflows: phones, scheduling, prior authorization, results follow-up, billing, collections, and EMR-adjacent administrative work. Plena’s thesis is that AI can run more of this operational stack end-to-end while working inside the practice’s existing systems. We liked the vertical depth, the wedge into painful back-office workflows, and the potential to become a deeply embedded operating layer for specialty care. Healthcare AI is crowded, but practice operations remain brutally inefficient and highly valuable if solved.

RASPIRE

What they do: RASPIRE builds runtime application security for mobile apps, protecting compiled applications from tampering, exploitation, and reverse engineering.
YC Profile: https://www.ycombinator.com/companies/raspire

Why we invested: Mobile apps are increasingly critical infrastructure for fintech, gaming, consumer software, health, and enterprise workflows. Yet many applications remain vulnerable once they are running in the wild. RASPIRE’s wedge is runtime protection: defending the app while it executes, not merely scanning code before release. We liked the severity of the security problem, the developer-facing adoption path, and the potential for RASPIRE to become a key protection layer for high-risk mobile applications. In a world of more AI-generated code and more automated attacks, runtime protection should become more important, not less.

Sherpa

What they do: Sherpa helps companies improve website conversion by using AI to test, optimize, and personalize growth experiences.

YC Profile: https://www.ycombinator.com/companies/sherpa

Why we invested: Growth teams know their websites leak revenue, but most conversion-rate optimization is slow, manual, and dependent on limited testing bandwidth. Sherpa’s opportunity is to turn CRO into an AI-native workflow: identify friction, propose improvements, run experiments, and compound learning across pages and customers. We liked the clear ROI, low-friction installation, and obvious pain point. The key question is whether Sherpa becomes a durable growth control plane rather than a point solution, but the wedge is strong and the buyer pain is easy to understand.

Superlog

What they do: Superlog is building autonomous observability for software teams.
YC Profile: https://www.ycombinator.com/companies/superlog

Why we invested: Observability has become essential, but it is often expensive, noisy, and painful to configure. Developers do not want more dashboards; they want systems that understand what is happening, collect the right context automatically, and help resolve issues faster. Superlog’s thesis is that observability should become more autonomous: less manual instrumentation, less configuration burden, more agentic debugging and system understanding. We liked the founder urgency, the technical wedge, and the category timing as AI changes how software is built, monitored, and repaired.

Userlens

What they do: Userlens builds AI customer-success agents that detect churn risk, generate account insights, prepare QBR materials, and help teams manage renewals and expansion.
YC Profile: https://www.ycombinator.com/companies/userlens

Why we invested: For B2B SaaS companies, retention and expansion are existential. The problem is that churn signals are scattered across product usage, CRM notes, billing, support tickets, customer conversations, and CSM intuition. Userlens turns that fragmented data into an AI CSM that can monitor accounts, identify risk, prepare playbooks, and support renewal workflows. We liked the repeat-founder angle, the clear pain from the founders’ prior company, and the possibility that customer success becomes increasingly agentic. The best version of Userlens is not a dashboard. It is an operating layer for revenue retention.

Voquill

What they do: Voquill is building an AI coworker for pathologists, starting with voice-driven report generation and expanding toward broader lab workflow automation.
YC Profile: https://www.ycombinator.com/companies/voquill

Why we invested: Pathologists spend an enormous amount of time documenting cases, producing reports, and navigating administrative friction. Voquill’s wedge is highly specific: listen as a pathologist works, learn their reporting style, and generate sign-out-ready reports. We like vertical AI products that start in a painful, repetitive workflow and expand into a larger system of action. Voquill is not a generic medical scribe; it is focused on pathology, where workflow specificity, reimbursement, and lab operations matter. If the company becomes embedded in pathology labs, the expansion opportunity is significant.

Zibra Labs

What they do: Zibra Labs builds high-performance computing infrastructure for quant trading firms running large-scale backtesting.

YC Profile: https://www.ycombinator.com/companies/zibra-labs

Why we invested: AI is accelerating the generation of candidate trading strategies, but backtesting infrastructure is becoming the bottleneck. Quant teams can produce more hypotheses than their systems can evaluate. Zibra Labs is building HPC infrastructure to make large-scale backtesting faster, cheaper, and easier to run across cloud resources. We liked the founder-market fit and the technical credibility of the team. This is a narrower market than generic cloud infrastructure, but it is a high-value buyer segment where performance, scale, and cost efficiency matter enormously.

Why This Matters for Fund III

This batch is a good example of why Team Ignite exists.

YC batches are now large, noisy, and fast-moving. The best companies do not wait politely for every investor to finish diligence. Rounds form quickly. Allocation disappears quickly. Consensus is usually late.

Our Fund III strategy is built around that reality: review the batch early, process signal quickly, meet founders before Demo Day when possible, and invest with disciplined speed.

Team Ignite Fund III is our second YC-focused fund. We have now made 101 YC investments to date in the fund, and the portfolio is already marked up at nearly 3x on 2024 investments. The fund is still young and still actively deploying, but the early signal is strong.

Fund III Snapshot

• Updated deck: https://tr.ee/yc_fund

• Join the fund: https://teamignite.decilehub.com/pacts?pid=7Na3Qz8D

• Closes End of June 2026

Our view is simple: if you want exposure to the next generation of YC companies, the edge is not waiting until everyone agrees. The edge is seeing enough of the batch, moving early, and being useful enough that founders want you on the cap table.

That is what Team Ignite is built to do.

Cursor was in talks to raise at fifty billion. Then a different conversation arrived.

Four days after SpaceX rang the Nasdaq bell on June 12, the company filed an 8-K, the regulatory disclosure public companies use to announce material events, to acquire Anysphere, the parent of the Cursor AI coding tool, in an all-stock deal valued at $60 billion. The fundraise turned into a merger. The price went up. The currency was stock that had been publicly tradable for four days.

Salesforce signed a $3.6 billion definitive agreement to buy Fin the same week, giving its Agentforce AI agent product a best-of-breed customer service layer. AWS turned bot payments and agent search into edge primitives, two unrelated-looking features that together outline a metered, cloud-controlled agent web. OpenAI launched a $150 million partner network with a target of 300,000 certified consultants by year-end and released a tool that simulates how a model will behave on real user conversations before it ships. The Fed held rates at 3.5 to 3.75 percent on June 17, with the FOMC statement saying inflation remains above the 2 percent goal. Translation: capital stays expensive while compute, power, and grid capacity stay strained.

Read each of these alone and they look like routine industry items. Read them together and the shape of the AI stack changes.

The independent application company, the place where most application-layer venture dollars have lived for three years, started to lose its standalone character last week. The middle is getting eaten from above by platforms turning features into primitives. It is getting eaten from across by strategic acquirers spending newly minted public stock. The squeeze is happening against a rate backdrop that is not going to bail anyone out.

The simplest test is to look at what got paid for.

What buyers paid for last week

Two of the better venture rounds of the week did not use the word copilot.

Convey raised $38 million in a Series A led by Andreessen Horowitz on June 17. The product is what the team calls operator-managed digital teammates: agents that nontechnical business operators build inside finance, accounting, marketing, and ad operations, that connect to legacy systems through IT-configured permissions, do measurable work, and write back into systems of record.

Gradial closed a $65 million Series C the same day, framing enterprise marketing as an execution problem. Their agents do the unglamorous middle of campaigns. Approvals. Publishing. QA. Reporting. The pitch is hours saved that show up on a budget line, not ideation that shows up on a slide.

Both companies look dull next to a foundation model release. Both are doing the most important thing a venture-stage AI company can do right now, which is own a workflow end-to-end. Write permissions. Audit trails. Integrations that take three months to put in and never get unwound.

Now look at the two acquisitions.

Customer service models are not scarce. Salesforce bought Fin for the route into its install base and the credibility of a best-of-breed agent already deployed at thousands of companies. Salesforce told the market its own agent platform, Agentforce, had hit $1.2 billion in annualized revenue last quarter, up around two hundred percent year over year. Adding Fin on top is a way to take a category that was rapidly commoditizing and lock the upgrade path inside an existing system of record.

The SpaceX end of the spectrum runs the same logic at a different scale. Cursor uses Claude, GPT, and an internal model called Composer. SpaceX bought it because Cursor sits inside a reported sixty-four percent of the Fortune 500’s developer workflows. That number is the company’s own and should be read as a marketing claim until diligence proves otherwise. Even if it is half right, the structural point holds. Developer tooling that has earned daily use across most large engineering organizations is a distribution asset that public stock can convert into ownership.

Two transactions in the same week valued at $60 billion and $3.6 billion. Both were about workflow real estate. Neither was about the model.

The other side of the squeeze

If strategics are buying the application layer where it stands on durable workflow, platforms are eating it where it doesn’t.

AWS announced two things last week that look unrelated and aren’t. On June 15, AWS WAF, the application firewall that fronts most large AWS-hosted properties, added a feature that lets content owners charge AI bots and agents per request at the network edge, with prices set by content path, bot category, or verification tier. On June 17, Amazon Bedrock AgentCore got Web Search as a managed tool. An agent built on AWS can now be grounded in current web data without wiring up a separate search vendor.

Read them together and the agent web has a price column and a default retrieval path. Retrieval goes through Bedrock. The price column lives at the edge of every AWS-fronted publisher. A startup whose product is web search for agents or a clearinghouse for bot payments saw its surface area shrink. A startup with a vertical knowledge corpus whose rights it controls saw it grow.

OpenAI’s two moves on June 16 echo the same logic. Deployment Simulation, in OpenAI’s own framing, lets the company replay prior conversations against a candidate model before release to forecast undesired behavior, surface novel misalignment, and reduce evaluation awareness. The Partner Network puts $150 million into systems integrators, consultancies, technology firms, and data partners, with a target of 300,000 certified consultants by the end of 2026. OpenAI is moving model release toward production simulation and enterprise deployment toward a partner-led channel. The shadow each casts over the venture market is the same. Independent AI eval startups whose product is a static leaderboard got harder to fund. Generic AI transformation consultancies got harder to fund. Tooling that plugs into the OpenAI channel and measures outcomes for partners got easier.

Microsoft posted enterprise AI guidance this week with the framing intelligence plus trust, model diversity, governance, observability, security, and FinOps for agents. Buyers should not be locked into a single model or harness, and should manage agent spend and behavior from one central control plane. Build something that strengthens that control plane and there is a place to stand. Build something that asks the enterprise to adopt a parallel system of record for agents and the sales motion gets long.

A macro that does not give the middle time

The Fed held the federal funds rate at 3.5 to 3.75 percent on June 17, with the FOMC, the Fed’s rate-setting committee, saying inflation remains above the 2 percent goal even as productivity growth and capital investment look strong. No signal of relief.

This matters more than the rate level itself. Capital stays expensive while compute, data centers, and grid capacity get more expensive. The companies that survive this environment are the ones whose unit economics work today, not the ones whose pitch deck has a hockey stick predicated on a model-cost crash that has not arrived.

For a wrapper company, this is the most painful macro available. They are already exposed to platform feature absorption from above. They do not get the rate cut that would buy another year of runway. The clock is the same color and it ticks faster.

DeepSeek closed a $7.4 billion fundraise the same week, reported by The Information on June 16, in a founder-controlled limited-partnership structure: no voting rights for commercial backers, five-year lockup, only the Chinese state AI fund taking direct equity. The deal mechanics are the news. Capital wants in. Founder Liang Wenfeng wants control. The signal for application-layer companies anywhere is that open-weight cost compression now has institutional patience behind it. The cheap end of the token market is going to keep getting cheaper on a schedule the closed labs cannot match.

Where capital still looks comfortable

Two places last week.

First, anything physical. Odyssey, the world-model company founded by ex-Wayve and ex-Cruise leaders, raised $310 million at a $1.45 billion valuation from a backer list that reads like sovereign and strategic capital playing one hand: Amazon, AMD Ventures, GV, EQT, In-Q-Tel. World-model capability claims remain vendor-stated until reproduced. Treat the headline cautiously. The investor mix is the tell. When strategics and government-linked vehicles anchor a Series B, the company is being underwritten as infrastructure.

Anthropic’s Frontier Red Team published Project Fetch Phase Two on June 18, showing Claude Opus 4.7 running robot-dog tasks roughly twenty times faster than the fastest human team had a year ago, with about ten times less code. The honest qualifier from Anthropic’s own write-up: the model still cannot precisely move a beach ball with the robot. The capability is real and incomplete. The investable claim is that fleet-learning robotics with a real data loop are getting cheaper to build per task completed. The advantage lives in the data, not the chassis.

Pegasus Tech Ventures and CYBERDYNE launched a roughly $60 million corporate venture fund on June 16 aimed at physical AI, automation, intelligent systems, and healthcare. The healthcare tilt makes this fund mixed-signal for a firm with an FDA exclusion. The broader pattern holds. Corporates with physical-world deployment channels want structured access to robotics startups, and they are willing to put a vehicle on the table to get it.

Second, anything that helps a buyer keep an agent on a leash. Agent permissions, agent identity, audit logs, deployment simulation, change management for nontechnical operators, model routing, FinOps for agents. The language is dry. The categories are the load-bearing plumbing of an industry whose buyers just got told by every major platform to make sure their agents are governed, observable, and secure.

The regulatory layer started taking shape

On June 17, OpenAI’s Sam Altman, Anthropic’s Dario Amodei, and Google DeepMind’s Demis Hassabis sat with G7 heads of state at a working lunch in Évian-les-Bains, France. Per Semafor, Altman pitched an international standards forum. Per CNBC, Amodei pushed a US-led coalition that excluded China from chip and frontier-model trade.

The substance of what was agreed is opaque. The structure is the news. AI governance moved from working groups to the leaders’ table inside one calendar week.

Pair that with reporting earlier in the month that Commerce Department export controls forced Anthropic to disable two frontier models, and with Politico’s June 18 reporting that the White House and Anthropic are now drafting a joint framework to assess AI security risks. The joint framework, when it surfaces, will be one of the most important regulatory documents in venture for the next two years. It will set the trigger for when a government can switch off a frontier model. Anyone evaluating an AI company today is also evaluating regulatory risk to its model access. That is new.

Questions worth asking the founders we are seeing this month

Last week’s pressures, taken together, point at a small set of sharp questions. The most important one is what Salesforce-Fin and SpaceX-Cursor answer in opposite directions: what about a company cannot be acquired or bundled away?

A few sharper versions:

• Which workflow do you own end-to-end, and which budget line at the customer disappears if you do?

• If Salesforce, AWS, OpenAI, or Microsoft makes your core feature a free primitive next quarter, what about your install base and your data stays defensible?

• Which single model are you locked to, and how fast can you fail over if it goes offline by government order?

• How sensitive is your gross margin to a forty percent drop in open-weight inference cost over the next twelve months?

• If your unit economics need a rate cut to work, is this a venture investment or a real-estate bet?

A founder who walks into a check-cutting conversation with crisp answers to those is in a small group. The rest will hear it from acquirers and platforms instead, on terms set by the buyer.

What to monitor over the next one to four weeks

A short list of items where the next signal will move underwriting more than last week’s headlines did:

• The White House-Anthropic AI security framework, when it surfaces. The trigger language is the document.

• Whether OpenAI’s confidential S-1 converts to public disclosure. Audited comparables would reprice every late-stage AI position.

• SpaceX-Cursor regulatory review timing. Closing risk is the most concrete pricing input on SpaceX equity for the next two quarters.

• Whether the AWS WAF AI traffic monetization is adopted by any major publisher and paid by any major crawler. If yes, content rights become a programmable market and a new layer of startups becomes investable. If no, it stays a feature switch with no economy on top.

• The conversion rate of robotics megadeal headlines into deployed fleets. Capital has been generous to physical AI for several quarters. The fleet sizes need to start showing up.

Closing thought

It is easy to read last week as a story about acquisitions. Two big deals. A handful of platform launches. A funding board that paid for operating loops and ignored copilots. A Fed that did not blink.

The structural read is different. The middle of the AI stack, the place where most application-layer venture money has gone for three years, started to lose its standalone character. From above, platforms are absorbing features. From across, strategic acquirers are buying workflow real estate with public stock. From below, open-weight cost compression has institutional capital behind it now.

What remains is a barbell. Infrastructure on one end. Companies that own a measurable workflow on the other. The middle is where most of the noise has lived. The middle is also where most of the next twelve months of portfolio markdowns will live.

The founders who already understood this were not on the news pages last week. They were heads down, signing into the systems their customers cannot live without, writing back into data other companies do not have, and learning from feedback loops that compound. That is the work.

As General Partner at Antler, Jeff co-leads the firm’s US Fund from New York, investing at what he calls the “inception” stage: before a polished pitch deck, before obvious traction, before the market has voted. Antler’s model is built around a simple but difficult premise: spend time with founders in person, understand how they think, watch how they operate, and back the ones who seem capable of building something massive from zero.

Jeff calls it “backing maniacs.”

Not reckless founders. Not loud founders. Not people performing ambition for investors.

The kind of maniac Jeff is looking for is someone with deep obsession, extreme urgency, resilient optimism, and a personal relationship with the problem they are solving. Someone who does not just want to start a company, but almost cannot imagine doing anything else.

That distinction sits at the heart of this conversation.

Subscribe now

From LinkedIn Hypergrowth to Day-Zero Venture

Before joining Antler, Jeff spent nine years at LinkedIn during one of the company’s defining growth periods. He held nine roles across sales and leadership as LinkedIn scaled from roughly 1,500 people to around 20,000.

That experience shaped the way he thinks about company building.

Jeff points to LinkedIn’s leadership, culture, communication, and focus as major lessons. The best leaders he observed did not simply tell people what to do. They taught people how to think. They gave teams frameworks, principles, and operating systems that created leverage across the organization.

That matters in venture because founders are not just building products. They are building cultures, decision-making systems, and teams that must survive chaos.

For Jeff, the LinkedIn years gave him a front-row seat to what high-performance organizations look like before they become obvious from the outside. That now informs how he evaluates founders at the earliest stage.

Why Antler Invests Before the Noise Starts

Antler is not trying to be a traditional accelerator. Jeff frames the firm as an inception-stage investor: a place where founders can answer the questions that come before the company.

Should I start this?
Who should my co-founder be?
Is this the right market?
What problem do I understand better than everyone else?
Am I actually ready to live this life?

Antler operates in 27 cities globally and has backed roughly 1,900 portfolio companies. The firm brings founders together in person through a residency-style model before equity or capital changes hands. That gives founders a chance to meet co-founders, test ideas, sharpen conviction, and decide whether this is really the arena they want to compete in.

Jeff believes this stage is less noisy than later pre-seed or seed investing. Once there is a product, a few customers, and early revenue, investors can get distracted by signals that may or may not matter. A customer loves it. Another hates it. A non-customer says they would never buy it. Suddenly the investor is buried in conflicting information.

At inception, Jeff is focused on the person.

Can they sell?
Can they recruit?
Can they learn fast?
Can they move with urgency?
Can they attract people into their orbit?
Do they understand something deeply?
Are they resiliently optimistic?
Are they unusually spiky in at least one important way?

That is the core bet.

The Founder Who Talks Too Much About Money

One of Jeff’s sharper views is that founders who lead with “I want to build a billion-dollar company” can make him nervous.

At first, that sounds counterintuitive. Venture capital depends on massive outcomes. Shouldn’t VCs want founders with billion-dollar ambition?

Jeff’s point is more precise: if a founder talks about a billion-dollar company too early, it can reveal that they already have a price in mind. If they would sell at $100 million or $400 million, that may be a perfectly rational life decision—but it may not fit the venture model.

The founders Jeff trusts more are often less obsessed with the financial endpoint and more obsessed with the problem. They may understand the economics. They may be realistic about acquisition offers. But their relationship with the problem is deepening over time, not weakening.

The rarest founders are the ones doing their life’s work. They are not building because a market map looks attractive. They are building because the problem has grabbed them by the throat.

That is the magic Jeff is hunting.

Why Antler Increased Its Check Size

Antler’s US fund is now writing checks of up to $600K at inception. That is a meaningful jump for a stage where many companies may not yet have meaningful revenue or even a finalized product.

Jeff’s explanation is straightforward: the quality of founders is rising, and great founders can do more with capital than ever before.

AI and modern software tools have collapsed the cost of building. A small team can now prototype, launch, and iterate faster than at any prior point in startup history. The right founders can turn early capital into real leverage.

But Jeff also sees a dangerous funding gap. Too many companies get stuck between not having enough traction to raise and not having enough money to get traction. Antler’s larger check is designed to break that paradox.

The firm typically splits the capital into two parts: an upfront check that gives founders room to build, and a follow-on check intended to help catalyze the next round. The goal is to let founders spend more time building the company and less time endlessly fundraising.

The Most Overrated Metric at Pre-Seed

Jeff’s answer is blunt: customers and revenue.

That does not mean customers are irrelevant. It means that at the earliest stages, traction is often noisier than investors want to admit.

A founder might have five customers, but those customers may represent five completely different stories. One loves the product. One is lukewarm. One hates it. One bought because of the founder’s personal network. Another may churn in two months.

At that stage, the investor can confuse motion for signal.

Jeff gives the example of Harper, an Antler portfolio company that later became one of the standout companies in the fund. When Antler first invested, the company had a different name and was pursuing a different idea. Jeff did not love the original concept. But he believed deeply in the founders, especially their intensity, speed, and ability to build.

The company pivoted into insurance and took off.

Had Jeff evaluated only the initial idea or early traction, he might have missed it.

That is the danger of over-indexing on metrics too early. At inception, the founder may matter more than the first version of the business.

The Case for Diversification at Inception

Jeff is also direct about portfolio construction. At the earliest stage, he believes diversification is not a weakness. It is a requirement.

If you are investing before the company is fully formed, you are taking extreme risk. The way to manage that risk is not by pretending you can perfectly predict winners. It is by building a system that gives you exposure to enough exceptional founders while maintaining the ability to follow on when the winners begin to emerge.

That is the logic behind Antler’s global model.

The firm sees an enormous top-of-funnel: around 150,000 global applications per year, with roughly 400 investments. In the US, Antler sees around 15,000 applications and makes roughly 60 to 70 investments.

But Jeff is clear that diversification alone is not enough. A great inception fund also needs a strategy to concentrate capital over time. You need the sourcing engine, the selection engine, and then the follow-on engine.

That combination—broad exposure early, concentrated capital later—is where the model becomes powerful.

AI Is Changing Startups, But Not Everything Becomes Software

Jeff acknowledges that AI is collapsing the cost of company creation. Founders can now spin up products, automate workflows, and launch businesses with far less capital than before.

But he pushes back on the idea that this means venture capital is dead or that all major companies will be built by one-person teams.

Yes, AI will create enormous leverage.
Yes, more founders will reach revenue without raising.
Yes, some software businesses will require fewer people and less capital.

But Jeff argues that the world is still full of hard problems that require teams, systems, complexity, and capital. Data centers, satellites, longevity, quantum computing, stablecoin infrastructure, physical-world systems—these are not trivial products that can be one-line-prompted into existence.

The future is not just “AI makes startups cheaper.”

The future is that ambitious founders can now attack bigger problems with more leverage.

Why Being Different Is Not Optional

One of Jeff’s strongest beliefs is that to be better than average, you have to be different.

That applies to founders. It applies to investors. It applies to how people pitch, build, hire, and distribute.

Too many founders show up with the same deck, the same language, the same AI-generated polish, and the same safe narrative. Jeff sees that as a problem. If everyone looks the same, sounding competent is not enough.

Founders need to understand where they are different and exploit that difference. Their wedge, insight, personality, obsession, speed, or worldview has to stand out.

Venture is not a game designed to reward average behavior. It rewards outliers. So the founder’s job is not to look like the median fundable company. It is to make investors feel that they are seeing something rare.

The Cave Walls Test

Toward the end of the conversation, Jeff shares a story from his managing partner about “the cave walls.”

The idea is simple: life is like leaving a cave, collecting experiences, and returning with stories, memories, and symbols that end up on the walls. At the end, the question is not just what you achieved. It is what you want to see on those walls when you look back.

Jeff uses that lens when thinking about where to spend his time and which founders to back.

Would he be proud of this investment in 15 or 20 years?
Is this founder building something worthy?
Does this company belong on the cave walls?

It is an unusually existential frame for venture capital, but it fits the way Jeff thinks about the job. The best investors are not just underwriting markets. They are deciding which people and problems deserve their finite attention.

The Real Job: Back Great People Before Everyone Else Understands Why

This episode is ultimately about what happens before the world has evidence.

Before the revenue.
Before the Series A.
Before the TechCrunch headline.
Before the perfect pitch deck.
Before consensus.

Jeffrey Becker’s job at Antler is to identify the founder before the market knows what to do with them.

That requires judgment, pattern recognition, emotional discipline, and a willingness to look past the obvious signals. It also requires comfort with being early, wrong, and occasionally mocked by the data that later becomes obvious.

The best founders rarely arrive fully legible. Sometimes they look too intense. Too early. Too weird. Too obsessed. Too different.

Jeff’s bet is that those are not bugs.

They may be the signal.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

00:01 - Introducing Jeffrey Becker and Antler’s Day-Zero Model

00:57 - Jeff’s Origin Story: Competition, Sales, LinkedIn, and Angel Investing

03:48 - Backing Maniacs at Inception

04:33 - Lessons from LinkedIn’s Hypergrowth Era

05:58 - Why Jeff Tells People Not to Become VCs

08:22 - How Antler Works Before a Company Exists

10:54 - Why Antler Increased Its Check Size to $600K

12:49 - How Antler Differs from YC and Traditional Accelerators

14:52 - Antler’s Global Founder Funnel and Selection Process

16:05 - How Founders Can Stand Out in an AI-Generated Pitch World

18:43 - Why “I Want to Build a Billion-Dollar Company” Can Be a Red Flag

21:57 - The Magic of Founders Doing Their Life’s Work

23:00 - Risk, Diversification, and the Math of Inception Investing

24:30 - Why More Early-Stage Bets Can Improve Venture Outcomes

26:40 - Using SPVs and Follow-On Capital to Double Down on Winners

29:03 - The LP Retreat and the Future of Emerging Managers

30:56 - How AI Is Collapsing the Cost of Building Startups

32:44 - Agentic Company Builders and the Limits of AI-Generated Startups

34:24 - Jeff’s Content Engine: Substack, Podcasts, and AI Workflows

36:35 - The Hidden Risk of Overfunding and High Valuations

38:42 - Boards, Governance, and Staying Aligned with Founders

40:31 - Antler Founders Who Redefined What a Maniac Looks Like

43:12 - Why Meeting Great Founders Keeps VCs in the Game

45:06 - The Sharpest Writing in Venture Today

46:26 - The Best Advice Jeff Lives By: Be Different to Be Better

47:48 - A Cold Intro That Turned Into a Standout Founder Bet

50:15 - The Most Overrated Metric in Pre-Seed Venture

53:14 - Why Jeff Changed His Mind on Valuation Discipline

54:18 - Breaking Rules to Avoid Missing Generational Founders

Transcript

Jeffrey Becker (00:00:00.131):

To be better than average, you have to be different. There is just no other way around that. I truly believe that if you obsess enough about things, you need to pick the things that you believe should be different and you need to exploit them. And I don’t think enough people look at the whole picture and try to find the exploitations. I don’t think they try to be different. I don’t have a lot of founders who change their pitch intentionally to stand out. I don’t have a lot of people that change the motion and the way they do things in an effort to stand out. I think that most people believe a good story, showing up and doing things when other people do it, is enough and it’s just enough. It’s just, it’s like, A, it’s boring and B, it’s like uninspiring, but it’s also just not enough. If you want any better than average, you gotta be different. And I think you gotta be different in order to stand out and you gotta stand out to raise capital and do attractiveness. That’s the thing I’m always challenging people to really think about and do and just lean into.

Brian Bell (00:01:09.782):

Hey, everyone, welcome back to the Ignite podcast today. We are delighted to have Jeff Becker on the mic. He is the general partner at Antler co leading the firm’s US fund out of New York. Antler is the day zero residency based pre seed model now operating in 27 cities with roughly roughly 1900 portfolio companies including some that you’ve probably heard of like Lovable, Aralo, Micro One, Pixverse and many others. Before Antler Jeff spent nine years at LinkedIn in sales leadership through its run to almost a billion members and I think like 20 over over 10 billion in revenue and so we’re delighted to have him. He writes the Monday morning meeting Substack and hosts Antler’s Further Faster podcast so I’m looking forward to learning how to do my podcast better. And so thanks for coming on, Jeff.

Jeffrey Becker (00:01:56.791):

Yeah, Brian, thanks for having me. Excited to dig into all things for you soon.

Brian Bell (00:02:00.353):

Yeah, so I’d love to get your origin story. What’s your background?

Jeffrey Becker (00:02:03.075):

Yeah, well, okay, so I grew up a little brother. I say that because I always talk to my founders about like, you know, what was it like growing up? Who were you? And I say that in part joking, but really because it made me very competitive. I was always competing with my older brother in sports and baseball and school. We went to the same college. How many years apart are you guys? We’re about three and a half years. So I just missed each other.

Brian Bell (00:02:23.422):

My boys are two years apart. And so I see this even more viscerally. Yeah. Now that they’re teenagers, they don’t even talk to each other. They don’t even like barely acknowledge each other.

Jeffrey Becker (00:02:32.464):

Yeah. My mom always said we were like, you’re destined to be best friends when you get older. And we always said, no, we beat the crap out of each other. But, you know, now we’re good friends. Yeah I mean that’s sort of like core to my DNA and I think you know being a baseball player you know being in sales really my career was always competing and trying to find ways to get an edge in something and I always felt like you know to be better than others you had to be different in some way and try to like find and really kind of elevate yourself against others you had to pick things that you could be different at. specifically especially in sales but then you know as I became angel investors I started a another company on the side as I worked at LinkedIn those are things that I felt gave me a bit of edge against others and so it’s always really kind of just about winning like drop me into any arena and I just want to compete on that thing but yeah LinkedIn as you mentioned is a great experience a rocket ship from you know 2012 to 2020 I did nine different jobs there over nine years basically all in sales but across three departments and lead our key accounts program for sales solutions before I left and then more recently angel investor founder of a company called Earhooks we kept earbuds in your ears for 10 years and sold those to a quarter million people around the world and then the center of all three of those things being at LinkedIn you know hyper growth and tech you know being a founder nights and weekends and scaling that thing globally and angel investing really to me was being a venture capitalist but you know I was never in finance and I hadn’t built my own SaaS business and so you know for me I really just had to go do the job and I started writing checks I wrote checks into you know loyal which just got you know their first two FDA clearances and are on the path to their third which would be the first company in history to, you know, sell anything FDA cleared for extending life of any kind. But the focus on dogs, which is super exciting one, Headspace, Misfits Market, Public Goods, Alto Pharmacy, just some really cool stuff that I got to angel into in the early days. And then now full time at Antler for four years, we’re, you know, really trying to do the same thing, which is back people and do that by getting to know them, spend time with them and underwrite the maniac and, you know, buy the basis of the future. and so yeah that’s that’s kind of me in a nutshell from you know personal life through to what I’ve been doing last 10 15 years that’s amazing so you guys

Brian Bell (00:04:45.298):

internally actually refer to it as picking maniacs that’s what I refer to it as you

Jeffrey Becker (00:04:49.540):

know I think there’s sometimes these words right when you say them other people resonate differently with and maniac is one of those like it’s just one of those things that I say that people seem to pick up on so Yeah, I call it backing maniacs at inception. My posts that you see on Substack are like screening for maniacs, you know, maniac mentality. It’s kind of like a little bit of a brand play, but I think it causes this like visceral reaction and some resonance with people on what it really takes to persevere, to hire great people to compete at a world class level, to have a deep relationship with a problem. And so it just sounds like it’s all-encompassing for something that is kind of esoteric.

Brian Bell (00:05:26.202):

I mean, you spent nine years at LinkedIn and probably had a lot of ups and downs there. What do you think you kind of take away from that as a VC now?

Jeffrey Becker (00:05:34.806):

It’s interesting. I mean, I think that at LinkedIn there’s a few superpowers of that organization, not the least of which is the leadership. I mean... I think you learn from people year round and having a front row seat to people like Jeff Wiener, Dan Shapiro, Mike Gamson, Shannon Stubo.

Brian Bell (00:05:49.544):

You were there kind of during the late golden age is what I might call it of LinkedIn.

Jeffrey Becker (00:05:53.869):

Yeah, right? Like post IPO, but we were there, you know, from 1,500 people to 20,000. Right. And seeing those people operate It really taught me a lot. I mean, it taught me a lot about culture and values and the importance of it. It taught me a lot about focus. Fewer things done better. It taught me a lot about communication. I think one of the real superpowers that those individuals I mentioned have is the ability to teach other people how to think as opposed to telling people what to do or, you know, which is kind of like a moment in time, kind of visceral, emotional thing in business often. Teaching people frameworks and the ways to go about decisions and the ways to operate in the confines of the culture and values I think creates an enormous, enormous amount of leverage. And I think that’s really overlooked. And I think people kind of gloss over, glaze past it. But when you’ve been inside of an organization that was that focused and operating that kind of way, I think it’s really clear that it’s the only way to do things.

Brian Bell (00:06:46.715):

Yeah. So you probably get a lot of young people in their 20s reaching out and being like, how do I become a VC? Right. And I remember like 10 or 15 years ago, I was like, how do I become a product manager? Yeah. What do you advise them now, given, you know, you can kind of look back at your career and kind of connect the dots?

Jeffrey Becker (00:07:03.376):

I mean, first, don’t do it. They absolutely definitely don’t do it. I just wrote a post, why you should never go into VC. It was like by far my best performing venture post. It kind of broke down the economics of it. It broke down the time horizons, the, you know, the math and potentially winning and actually making any money doing it. And the bottom line there was like, don’t do it unless... I can’t talk you out of it because you really just like anything you really have to be completely obsessed you have to be really giving it 150% if you want to compete on you know a real stage or a real level so I think first is don’t do it and you kind of see like does someone jump over that wall or not I think the second thing is if you do want to do it you should you should just go do the job right I think a lot of people try to figure out like almost like double dutch like when do I get in you know like waiting for the job interviewing and yeah yeah like talking to people when do I do this how do I go to that and the real answer is like go find deals write your own personal checks if you can’t write personal checks syndicate SPDs talk to the investors talk to the founders get out there and do the grind and like you know find out a if you like it b if you’re good at it you know see if the founders like you and appreciate you You know, so on and so forth. Actually, it’s advice that I got from Mike Amson, the chief revenue officer at LinkedIn, you know, back when I was there. I had an offer to leave LinkedIn at the time and he was like, what do you want to do? I always want to be a venture capitalist, which is another story which we can get into. Yeah. He was like, well, if that’s what you want to do, don’t take this job. Stay at LinkedIn. You clearly have control over it. You’re doing well. Get yourself promoted. Do whatever you need to do and take the free cash flow and go do the job and find out if you like it. And that’s what I did. It gave me a track record and a resume. that I just, it just was an excel, you know, of like the deals I had done and the things I’d gotten. And I used that to kind of get into VC rather than just trying to sell someone on my potential. It was like, here’s what I’ve done. You know, well, well done is better than well said.

Brian Bell (00:08:55.387):

Yeah, I love that. So let’s talk about Antler, you know, walk us through like how you guys operate what’s actually changed in the last couple years you know now you guys are up to I think 500k at day zero which is crazy you know before companies has any revenue or really a product why the job what’s new there and how you know how has the strategy changed

Jeffrey Becker (00:09:15.680):

yeah for those that don’t know antler is the world’s largest inception fund we have 27 offices around the world the thing that ties us all together and the way that we operate is working with founders in person before any equity or money changes hands so that those founders have a chance to meet co-founders work on their ideas not over commit to the cap table and to the endeavor before they even know if it’s something they want to do it’s such a such a hard hard hard life you know to go do and go live that way And so to spend a little bit of time around other people that are just as crazy as you, just as smart as you, and really find out if it’s the way you want to spend your time, if that’s the arena you want to compete in, it’s really important. Finding co-founders is super important, picking the right game to play, making sure that you are around people that are supportive and helpful as opposed to in your apartment trying to code something into existence. I think that is something that brings all Gantler locations together under a common thread. And by working with those people in person, you get to know them super well. What are their motivations? How do they operate? How do they sell? How do they think about the future? What do they know better than anybody else? How do they attract people, customers, investors, all these sort of things? And that is actually, in my opinion, a much less noisy stage to invest at compared to Six months later or seed where there’s a product, a customer, a team and, you know, there’s just a lot of things that can go right or go wrong. And it’s an emotional thing when you’re in that stage of investing. Whereas when you’re in the inception stage and you’re just talking to the people and understanding them, they are things that are just like absolutely have to be true now and in the future. And when you can identify those things and identify if they’re extremely spiky at them, I think it’s actually a really interesting place to write early stage checks and deploy capital. Typically, we’ll write a check in like two to four weeks. So you’d be in the office for a couple weeks. In some places around the world, you might stay as long as like two or three months. So, you know, people do it differently in different regions. In some regions, you might get 100 or 200K. As you mentioned here in the US we’re actually writing 600k now and the reason for that change is honestly like the founders are amazing and it’s very obvious to us that the quality just keeps getting better and better and they know what to do with the money they can create more leverage than ever before with the tools at their disposal and you know my personal belief is like if you have this much conviction in people Lighting them 200k versus 600k is the difference in them spending their time building the company and then spending the time fundraising. Because the previous kind of earlier version of this, what we found is that companies were stuck between not enough traction to raise and not enough money to get traction. kind of a paradox whereas this you know 600k we usually break it up into two checks usually it’s like 350 or 400k up front and then we give them an uncap check to the next round to help catalyze it and so it’s both enough money for like a year year and a half some team building some technology go get customers prove there’s you know real need for this and it’s a catalyzing check to get the next round. And for all those founders, we’re basically setting it up with something like 40 investors on average each for their seed round. So 75% of them are raising money within three to six months of our first check. So that’s kind of the headlines. We do have a Series A fund that follows into these thousands of companies around the world, which allows us to go all the way to IPO. We’ve deployed, you know, money all the way through to Series B and C. Oralo, our first unicorn, you know, we backed them, you know, seven years ago. I think they just had like a unbelievable, you know, record month that Bahadir just posted on LinkedIn, you know, some crazy number, tens of millions of users and Ibiza. And it’s just incredible. So we’ve been able to back them through every round, which I think most firms Operative Inception don’t typically do so yeah so a lot of founders listening would be

Brian Bell (00:12:58.796):

probably asking themselves, okay, how does this differ from like a YC Techstars kind of model where it’s a residency, you know, 12, 16 week kind of program accelerator. It sounds, it’s almost like similar in a way, but different.

Jeffrey Becker (00:13:12.026):

Yeah. I like to think that we’re at a bit of an earlier stage. The way I think about it is like, should I, shouldn’t I? You know, like, should I start this? Should I not? Who should my co-founder be? You know, what is the right way to start positioning this or who are the right customers? And that’s a really interesting time. because everyone goes through it right whether you’re like the best found in the world or or not everyone goes through that moment and if you are there in those moments it’s very revealing about how the founder will think about making decisions how optimistic they might be how creative they are how resilient they are how much clarity they can lock in on something and on an idea the slope and the speed at which they can move there’s a lot of information in that stage but if I look across at my peers and they’re I mean they’re phenomenal and I never have anything bad to say about folks I think there should be more and more firms doing this they really require a pitch deck a team you know clarity over what you’re building because they’re giving you the money right away and what we’re saying is like look maybe you should maybe you shouldn’t that’s okay it’s important that you’re around other maniacs you want to be around a ton of people and find a great team and go founders and figure out what that sixth gear is and if we can do that in two weeks or four weeks you’re going to be better off for it and then we’ll give you that money and then we’ll go do the exact same like let’s go get to a seed round as fast as makes sense for your business so we will also help people fundraise and run that kind of second phase which is akin to what like a 15-week sprint to a demo day would look like but we’re doing this first part that’s a bit unique and I think very additive to founders and culture as a product is a thing that I think goes back to those LinkedIn days we were talking about the culture is just it’s everything the team you build is the company you build and you know that’s what we’re trying to create around the founders the very beginning

Brian Bell (00:14:49.549):

Yeah, it’s really interesting because you guys are capturing, you know, the startup interest and helping right from inception, you know, from like, hey, like, I think I might want to do this, right? And you guys are only accepting something like 3% of those applicants. which is interesting so you have this huge funnel what does each stage of that funnel look like what are some of the milestones you’re looking at obviously when you’re going you’re picking zero you’re picking from zero that’s that’s a whole different kind of thing yeah it sounds like you know three percent you know go from hey I want to work with antler to I’m working with antler and I’m you know coming into the office and figuring all those things out and then there’s another step which is I think it’s something like ten percent actually get the the first kind of money in and then you obviously have the natural transition from there raising a full pre-seed or seed and then an A and everything but maybe you could walk us through a little bit for founders listening considering applying at the highest level forward how do they kind of tell you that they’re maniacs so that they they’re one of the three percent that gets selected for the for the

Jeffrey Becker (00:15:53.511):

residency yeah honestly the my tools on defining a maniac have to change because the presentation layers change like everyone’s pumping out a Claude deck you know and like honestly all the fonts are the same you can tell everything is AI generated so I’m updating my own system at least tell Claude to change the the font yeah change at least change the font yeah I love that at least change the font yeah You got to do something to stand out, right? I said that earlier, but like as founders, for sure, you want to be different. You got to stand out.

Brian Bell (00:16:19.308):

I’ll give you the numbers and I kind of get through some of the questions you asked me. I think at the highest level, 150,000 people apply every year globally.

Jeffrey Becker (00:16:25.874):

From that, about 4,000 people get, you know, a really serious look. And then from that, about 400 investments. That’s how you get from that number, you know, down to about, you know, 3%, which is then about half a percent of an investment. And that made us the most active investor in the world last year. The way we’re able to do that is we’re running these 27 cities. So we actually have 200 people. Globally, you can think about it like a platform with lots of venture firms sitting on it. Here in the US, those numbers are a bit different. We have three cities. We just launched San Francisco last year. we have New York and we have Austin Texas we get about 15,000 applications a year and we’ll make about 60 to 70 investments so that’s kind of our purview inside this

Brian Bell (00:17:05.412):

global because they kind of started I think in Canada right we actually started in Singapore yeah in Singapore and then it kind of spread out from there yeah yeah

Jeffrey Becker (00:17:13.362):

that’s right That’s kind of how the funnel works in terms of the numbers. But I think to your question on standing out as a maniac, like each of the partners has their own things that they’re looking for, right? Part of the value in this is that we’ve hired extremely smart people who’ve done incredible things, starting their own companies, building their own firms. And what they’re looking for is different. So I am one person in one brand of investor. I really like to understand people’s psychology, the hard things they’ve done. I like to know their relationships with a problem they’re working on, the reason for being and why they’re doing that. I also just want to see the sense of urgency and the slope. I want to understand how resiliently optimistic they are about the work that needs to get done. You got to be front loading that time. And so I typically try to meet people over three to five meetings. I will kind of plot those data points in my head of like, who are you psychology wise? You know, tell me about your obsession. Tell me about your execution. And let’s evolve that like week to week really, really quickly and find out if you spike on anything that is extraordinary. And if that’s the case, then I’m probably going to take it to an IC, do some research, build a business case around that founder and around that business and try to write that check as quickly as possible.

Brian Bell (00:18:21.151):

So in your 50 million lie essay, you said founders who lead with, I want to build a billion dollar company make you nervous. And the ones that barely talk about money are the ones you trust more. So this kind of contradicts the standard VC kind of pattern matching, right? So tell us more about that.

Jeffrey Becker (00:18:37.492):

Yeah, I think if you think about those two, you are talking about building a billion-dollar company, your intentions are to make money. So in my view, given the opportunity to sell that company, you already have a price on day one in your head. You better be talking about a trillion, not a billion, because if you have a price in your head at a billion and you get offered 400 million, my LPs and I, we’re not going to be building a 10X, 100X fund out of that. The ambition’s not high enough for someone that’s just talking about a billion dollars, especially if they have a price tag on day one. Their ego is already going to be a problem here.

Brian Bell (00:19:12.728):

There is an entire... I just look into a casual conversation sometimes with a founder. I’ll be like, what’s your number? If Google wanted to buy you guys for 100, would you sell? And I kind of do it in a very coy, casual way. I can sometimes fish it out of a founder. Like, yeah, we’d sell for 100. Yeah, definitely.

Jeffrey Becker (00:19:30.263):

Great, yeah.

Brian Bell (00:19:31.303):

Thanks for letting me know.

Jeffrey Becker (00:19:32.333):

yeah and that’s okay like venture capital is not for everyone there are tons of ways to finance your business and build your business and that’s not to say that I don’t appreciate earnestness and honesty and intellectual kind of you know appreciation for reality but the way you answer that question to me is important there are other tells other than like yeah I’d sell for 100 million someone could say to me hey yeah if someone offered me 100 million today it’d be kind of silly not to take it but my relationship with this problem is deepening I think there’s a trillion dollar opportunity here let me tell you why let me walk you through that I don’t know if it’s true right now, but there are some risks that I need to reduce or things that I need to figure out. And if I can figure those things out, it’s going to be harder and harder for me to sell in the future because of what I think we can do here. And I think that that’s a really honest way to go about that question. And it tells me something about the founder. It tells me something about how realistic they are, how well they understand the complexities of things. And then there’s this other side of the spectrum, right? Like you have the, I’ll take the money now. You have the people like, I would take the money because I built nothing yet and that would be insane. But the problem is hard and here’s what I’m doing. But then you also have these people that are just like doing their life’s work. Like, I don’t care how much money you give me. Like, I’m just going to keep going and building because this is the thing that I was put on this planet to do. And there’s just like so few of those people. yeah but when you find them it’s so fun to talk to them you know that’s magic yeah

Brian Bell (00:20:47.317):

yeah because it’s a little bit it’s all about the mission when it’s all about the mission for them it’s not about them yeah and it’s like jealousy right because

Jeffrey Becker (00:20:54.568):

you’re like I would love to feel that way and be like

Brian Bell (00:20:57.832):

That’s what most people don’t get about VCs is like how jealous we are when we found great founders. It’s not like, oh, I want to be you, but it’s like, I wish I felt like that about something. And maybe we feel like that about being a VC, but yeah, those are the best founders, right?

Jeffrey Becker (00:21:16.221):

For sure. For sure. I think when you see something so clearly, sometimes you have to go do it. And being a VC, you see how hard it is and you see how low the odds are. And actually, it’s funny, our managing partner here in the US on that happened recently he saw an opportunity that he was just like I cannot believe we’re not building this and he spun out last year and he’s just doing amazing building an unbelievable generational company and you know we’re excited to be his first backer so obviously that’s you know there’s not too much that’s public yet but it does happen to some of us we get the itch and the bug and and yeah I’m really excited to see what he builds

Brian Bell (00:21:47.965):

That’s really cool. Yeah, and I think I think this is like a secret. I love that story because I think it’s like a secret wish that all VCs have is to have something grab us so by the horns by the antlers to like and we just have to like go do that and shut down our venture for stop making investments and just go be a founder and work on something that drives us like that yeah so there’s there’s this thing in VC where you pass on things that you should have invested and you invested in things that you should not have how do you guys kind of adjust your process you know personally and at the organizational level and how do you avoid becoming too risk on or too risk off in that process

Jeffrey Becker (00:22:24.580):

That’s interesting. I don’t know if there’s something to risk on. I think you’ve got to be taking the risk. We’re trying to build trillion dollar enterprises, right? We’re not here to make a little bit of money. We’re not here to take part. We’re here to take over, right? As Conor McGregor would say. And I think to do that, you’ve got to be extremely ambitious and you’ve got to be thinking about how to create something generational. Because if you’re just trying to return two or three X, which is better than most VC firms, you’re better off being in the S&P 500, right? This is a part of an asset class, in my view, at least at this stage, can be an insurance policy and a lottery ticket at the same time. Makes the math very attractive. If you do enough investments at this stage, you’re going to get that two, three, four or five X pretty systematically. At least that’s what we’ve seen in the data. But you’re also going to have this lottery ticket of being so risked on that you expose yourself to the Anthropics, the Calsies, the Cursors, the Airbnbs of the world that had early stage finance things like this. And so we want to keep that risk on as much as possible. But the way we take the risk out is not by the investment itself. It’s by the diversification. It’s the number of investments we make. And that’s how we can balance those two.

Brian Bell (00:23:27.874):

And I love that so you’ve got completely into diversification it sounds like and this is something that I get into disagreements about with other VCs LPs especially as I’m raising my fund and because we’re very diversified pre-seed funds right we do 100 investments a year and I’m like just look at the math this is how like the math works this way I literally have a tool on my website it’s a Monte Carlo tool which with the probability distribution and you can actually assign all the different probabilities on the distribution and you know pretty much any assumptions you you grab from any data set cartas you know whoever it just the math says do more investments so it sounds like you’ve fully bought into that

Jeffrey Becker (00:24:06.814):

It’s yes and to that question. So a few things. One, I mean, look at SV Angel, look at YC, like just look at the people that have done diversification well. It’s storied levels of multiples on their phones. I think that when you do good diversification at this stage, you get high multiples. And I think that’s important. However, you also need to be able to concentrate the portfolio as you go. You need to find a way to deploy more capital to increase gross returns, right? And so, you know, you see that in YC continuity, you see that in a few other places. And we have a strategy for that here as well. And I don’t think that it’s one at the exclusion of the other. You have to build a world-class sourcing engine, a world-class way of making decisions, a world-class way of getting ownership and diversification. But then you can’t just leave it there. You also have to figure out what is a world-class way to follow on and actually create, you know, systematic value and major, major gross returns for your LPs.

Brian Bell (00:24:57.480):

Yeah, and it is to follow on. I’ve noticed this in my portfolio because I haven’t reserved capital yet. You’ve got to increase ownership. I mean, you guys do a really good job of this, right? Because you’re grabbing 10% of the company very early and then you’re following on to maintain that ownership. Something I struggle with as a small check writer. What I’ve noticed is... My SPV strategy, my follow-on, actually performs even better than the fund because I’m sort of picking the winners coming out of the fund and doubling down and increasing ownership through SPVs. So it’s something that I need to lean on probably more in my fund as I go along.

Jeffrey Becker (00:25:28.676):

I think you have asymmetric access to those deals and you’ve got asymmetric information. And if people want to be in those SPVs, they should be required to invest in your underlying inception funds and it’s the same thing you see at Sequoia right like everybody wants in the seed vehicle you know that’s where you got Instacart that’s where a lot of real returns land and so you got to participate in the rest of it too and you know if you deploy a billion dollars and get 3x you make amazing gross returns that’s very hard to do at inception to deploy a billion dollars at inception you’d have to write you know hundreds and hundreds and hundreds of checks right we do we’ve done 2,000 checks across 27 cities to do that and so the infrastructure you need to be across that much deal flow is you know it’s antler however I do think that like relative to your fund size you can get enough diversification you build up good systems to monitor that portfolio and stay close to those companies. Yeah, you have a right to win and you’ve got a source of deal flow. And so those STVs is a way to basically plow money into the best companies. And that’s where you can get gross returns. So you get your multiple here and you get your gross here. I think collectively, It’s a really powerful strategy, especially when you’re close to the founder. It was like, you know, you’re not only helping them, but they’re helping you. And in some way, like you’re building it together. I think that’s really part of the magic, too. Like if we do this for a few decades, we’ll look back and we’ll have a network of people that not only that we love and love us as humans because we did this thing together. but like it’ll just it’ll be more fun right and if you just give them that first check and never do anything else with them it gets quite complicated later and I think it’s important to be there for the long haul.

Brian Bell (00:27:02.092):

Yeah so you recently cited Dan Gray’s LP survey showing 57% of LPs went back emerging managers this year which I am which sucks up from 33% the year prior so you guys are now institutional what does that LP retreat mean for you guys and for the industry at large?

Jeffrey Becker (00:27:20.945):

Well, you know, look, I wrote that looking a little bit backwards. I’m not trying to project onto the market or like, you know, tell people how things are going to go. What do I know? But the last few years, LPs have retreated, right? Like there’s a big decline after the 22, 23 timeframe. I think we lost 74% of the LP capital into funds. We lost something like 60, 70% of the emerging managers evaporated. However, At that very same time, there is a huge swath of people that have just worked in a bunch of amazing hypergrowth companies over the last two or three years that are now going to spin out. They’ve become millionaires themselves, whether they worked at Anthropic or Lovable or wherever it might be. And they have an amazing network of engineers and other leaders that were exposed to that culture and that leadership and that hypergrowth. And I think those people will be well-suited and primed to bring in emerging manager money again and tell a really good story about asymmetric access, the AI narrative and the platform shift we’re seeing. And so I don’t know if what has happened will continue to happen, but certainly there’s a moment right now and every manager that I’ve spoken to is closing a fund, raising the next one. They’re all in that moment of like, can we capture this lightning in a bottle? And that’s another topic in itself, right? Are we capturing the token value? is it a bubble or not so lots of variables but it’s just fun it’s like the energy is back and that’s a fun time to be an adventure for that reason

Brian Bell (00:28:44.783):

Yeah, yeah, totally. Speaking of tokens, AI is collapsing the cost of building a company. So there’s a real argument that early stage funds are being disintermediated, right? Solo founders or duos do not need a ton of money to get going and get traction. I see this all the time, but we haven’t raised any money. We have a million of ARR, right? Now we’re starting to raise and that’s like, that creates a little bit of gap. How are you guys seeing that in the market? And how is, you know, this time the same but different?

Jeffrey Becker (00:29:12.408):

yeah I mean there’s like that chart I don’t know if you’ve seen it it’s like number of employees per million dollars of revenue or something and it’s just like going to zero it’s just like it’s very clearly on a trajectory towards like more and more and more efficiency and higher and higher leverage right yeah it’s I mean you can do more you and I could go to Lovable right now go to Antler Portfolio Company and type in I want to build a fund you know or a deck you know analyzer for founders it costs $10 sync up your Stripe payment gateway you know So load in an investor database, build a little algorithm with natural language, and founders could be paying you $10. You’ve got distribution for that. You’ve got the know-how on how to build that thing. You can do it with basically a keyboard and a screen. You can start a business. Is that a trillion dollar company? I don’t think so. I don’t know. But you can start a business. And I think that is really interesting because you can do a lot more, a lot less. You can execute in places where you may not have been able to last year or the year before. And those models and those businesses are the worst they’re ever going to be. So will we have agentic situations where you can just one line code of business into existence probably it seems like it but that’s also a very

Brian Bell (00:30:22.110):

online go like oh that that’s a trillion dollar company and just spin up like a swarm of 10,000 agents to build it yeah well there is a company that’s doing that

Jeffrey Becker (00:30:29.957):

there’s actually two I saw one that just joined YC and I saw one that got funded this week yeah I saw that one right it’s got like AI slot backwards like Pulscha or something So those things are coming and they’re cool. I think it’s interesting. Will the companies that they build be trillion dollar companies? No, I think that’s a lot more like Shopify, right? Where 95% of those people that start stores like don’t actually spend millions on advertising, driving traffic and revenue, but 5% of them create a ton of value for Shopify. So I think you will have company builders, agentic company builders like that, that build lots of long tail businesses. I think if you want to build a real business that is, you know, hundreds of millions in revenue, That’s still going to require people and complexity and structure and systems and teams and capital.

Brian Bell (00:31:10.328):

I also think it’s a really small kind of,

Jeffrey Becker (00:31:11.869):

not small, but like a myopic view on the market, right? Because in the same breath, we also have companies building data centers in space and launching satellites and fixing you know longevity and working on quantum computing things that are still you know atoms instead of bits and I just think that the news media would have you believe that like everything’s going to be AI and I think that yes AI will create a lot of leverage but we also still have like this massive physical world and a ton of value to be created there as well and so as generalists as people that get to work with you know 150,000 people applying every year get this massive data set on like what are all these smart people thinking about like what do they think the world’s going to look like and that’s just like a cool Yeah, that’s awesome.

Brian Bell (00:31:53.108):

Let’s talk about the content engine. You know, you’re writing a blog, you’re doing a podcast, you know, asking for a friend, like, what are some learnings from that? If you could start over again?

Jeffrey Becker (00:32:02.399):

Yeah, if I could start over again, I probably wouldn’t do it. It’s the same answer as VC. No, I mean, I think that it’s I’m the kind of person like when I commit to something I commit like I just failure is not an option it’s not like a thing that I understand or maybe my ego can’t deal with it but I just right and so I committed to writing this blog every Monday I’ve now done it for you know five or six years straight every single day every single Monday rather recently I built a content engine around it so it’s basically a set of MCPs that plug into things like granola slack email etc I try to record the calls I’m doing them and then I have a like a master prompt one for coming up with topics so try to figure out like what should I talk about where are their angles and things where is their data that supports you point of view and then another agent that I just kind of drop that context into where I edit and I work and I try to like revise my thinking and sort of do the mental tennis with the machine and then a third agent that does all the branding so I elevated the brand of theSubstack elevated like the content generation which you and I are doing all day every day talking to founders like why do I have to sit there and redo that on a Sunday or Monday morning when I can have the machine sort of ride shotgun with me and then the distribution of it is like how do you write LinkedIn posts or how do you write tweets or how do you do things that amplify and magnify the work that you’re doing so I’ve been getting a little bit better at it you know it’s sort of a compounding effort the graph kind of chunks its way up slowly but surely on the podcast actually I found a studio here in New York you just literally book it and show up they do all the editing they do all the stuff and so it’s actually quite a simple lift but again what we really want to do is bring great people into the studio and really just talk about Inception like talk about the stuff that no one talks about You can see the headlines of so-and-so does a $500 million round, so-and-so is worth a gazillion dollars. But there’s actually an inverse relationship with the amount of money people raise and the level of success. That’s a Dan G favorite, right?

Brian Bell (00:34:01.707):

Tell me more about that. What is the inverse relationship?

Jeffrey Becker (00:34:04.130):

It’s like the overfunded companies spend money in ways they shouldn’t. They lose focus and they die. Or they get overfunded and the valuation is too high to catch up with. Or the liquidation preferences are too much and they can’t outrun it. And it’s really just like a tale as old as time.

Brian Bell (00:34:17.228):

Sometimes you look at a company that gets, you know, they raise a bunch of money and what they don’t tell you is the reason they raised a bunch of money at such a sky-high valuation is there’s like a 2x Lickpref on the stack there, right? Which is just going to just cut them off at the knees. Yeah.

Jeffrey Becker (00:34:33.067):

And then you, you know, you bring these founders into the studio, ones that build unicorns, like we’ve had founders of Superhuman and Rent the Runway and, you know, Cameo and others. And, you know, they all tell you the same thing, right? Like the job of building a company is to find a customer and keep it. It’s not to have a high valuation, right? David Politis the founder of BetterCloud he runs a podcast called Not Another CEO he’s interviewed like hundreds of CEOs and he’s in there telling me the exact same story he’s like you need a valuation that gives you options right you don’t want a valuation that leaves you no option other than to build a you know a trillion dollar company because it’s unlikely the market might change the customer set might change there’s COVID might happen there’s so much that’s out of your control and so as a founder How do you remain in control? And how do you kind of remove the ego of the valuation and focus on finding and keeping a customer? And doing that at a rate and a speed that allows you to gain market share and just do that for an extremely long amount of time. And you’d be lucky to find a second or third act. Chesky talks about how Airbnb is trying to find their second act. just the greatest companies of all time yeah some of them still are on their first act you know and Amazon obviously found a second act and AWS and you know a third act but I think people underestimate that just like doing one thing extraordinarily well and you know ignoring some of the hype and valuation kind of ego game are you

Brian Bell (00:35:53.267):

on any boards do you kind of aspire to sit on boards does Antler sit on boards like how do you guys approach that

Jeffrey Becker (00:35:58.201):

I technically am on a couple, but I don’t aspire to be on boards. There’s reasons for the ones I am. But the reality is like, I just want the founders to be successful and I don’t want to be at odds with them. I want to make sure that we have, you know, major investor rights for our LPs and make sure that we have information rights and then the things that we need to manage the portfolio as opposed to just like, you know, YOLOing uncapped notes into, you know, AI companies and crypto companies when things are hot. I want to make sure we have the right management in place and the right sort of governance. But on the board thing, you know, I’m happy to do it if founders want me there but really what I’m because I’m coming in so early what’s good for me is good for the founders and vice versa right like if they get to a place where this is like no longer their life’s work and they want to sell it at you know 100 200 million dollars or whatever the number is you know sure take the check like we have the same economics here we’re both starting from basically zero together and if you don’t want to sell it then don’t and I feel like being on the board it just creates a bunch of other paradigms I’m also not an expert in any specific industry. You know, I sold for a long time and there are things that I know well, but I think the founders are better off having people that are like just absolutely legendary and have done that thing for decades. And they can really make sure that those companies succeed. So I think that relative to our position on the cap stack, we share very similar incentives. And also there are people that are better suited most likely or most often to sit in that seat and guide the founders. So I’m happy to just kind of be there as a friend and, you know, give them the real talk and, you know, be the person that they can call when, you know, they’re missing payroll or can’t raise a round or, you know, do the founder therapy thing because that’s the moments when their relationship’s really built, you know?

Brian Bell (00:37:34.940):

Yeah, I love that. Let’s wrap up with some rapid fire questions. Which Antler founder has changed your mind about what a maniac looks like and how?

Jeffrey Becker (00:37:43.644):

Oh my God, there’s so many. There’s so many founders that have just taught me so much. Like, I just to rattle off a few like Casper Barnes from Amino Chain has showed me what life’s purpose is about and really swinging big I mean it’s an unbelievable company you should look into it it’s just one of my favorites I have a founder who’s building in Africa right now who I swore I wasn’t going to invest in Taylor Rowan from Honeyguide and he’s showed me how a quant mind can really manufacture economics inside of a business that are just like undisputed and it’s unbelievable what he’s been able to do on a business that maybe seems less than interesting from the outside but then once you start to get in there and see the spreadsheets and the numbers the idea of manufacturing just like incredible amounts of value it’s really made me appreciate the quant kind of mine and Casper Life’s work if you meet Shosh from Doorstep this is someone who just is magical like I really believe that like everybody is going to know this guy in our futures he’s just someone who if you spend 15 minutes with this guy the level of magic the level of optimism the way that he speaks the way that he the way he appreciates the things that you’re supposed to appreciate and ignore the things that you’re supposed to ignore I just have never seen it in someone his age he’s so young he’s so full of raw talent if you talk to Arthur from Argentio he is just absolutely obsessed with the quality of his team and the culture and the types of people they hire and how important important recruiting is and so like as I’m exposed to these different people you sort of like manufacture this ideal person in your head of like who is the can I take this person this person this person and put them together but there’s no right way to build a company right like that if there were a playbook people would be coding that into an agent and building unicorns and what makes these people interesting is that they’re all so different and they’re all outliers in different ways and that superpower is what they’re leaning into the most to build their companies you know or like Dakota from Harper you know he’s like a darling adventure right now in the valley emergencies led you know series a is really storied round I mean that guy is up at 5 a.m in the office and I don’t know if he ever sleeps he’s always like an instant responder and he just is so switched on and his sense of urgency is so so high and I think it’s just it’s really incredible when you’re exposed to these people and what you learn about world-class what it looks like yeah I love that and I love

Brian Bell (00:40:01.027):

that response because it just reminds me why I like being a VC it’s meeting really amazing founders and I’ve written some articles and tweets on it it’s just like you know sometimes you don’t want to come to the office you’re just like I just want to like stay in bed and like you know not go to work and then you go to work and you just meet this amazing founder you’re like oh this is why I do this you know

Jeffrey Becker (00:40:18.892):

Yeah. You sit down, like I was just sitting with Zach from HiFi and we’re just talking about this like quadrillion dollar token problem. You know, Ramp just raised at 44 billion and they talked about this idea that like the ROI and the measurement of AI and the tokens that are being consumed is going to be like the crux of the next 10 years. And how do you measure that? If you could tokenize your spend, and understand almost like a stablecoin infrastructure or tokenized infrastructure. If you could really understand the ROI the way we understand the ROI of advertising dollars, for example, like the reason Google is Google, I mean, it’s going to be a mega trillion dollar industry. And, you know, HiFi is building these APIs for stablecoin and it’s unbelievable what their customers are doing. just sitting with him and talking to him about some of the use cases. It’s like, even if it ended today, we have moved so much money around the world in a way that was never possible. And you’re just like, wow, by some extension, or Redditus, which just raised their Series A, they went to YC after us, and they’re going to take this satellite into space and do research in microgravity. And it’s just like the level of ambition of these people and the things that they’re going to accomplish and the things that are going to develop you know it’s I always feel lucky to be a VC and be like a small part of that because I’m not qualified to do any of those things but by virtue of doing what we do we get to be the people that invest in them and believe in them and give them that shout to realize their potential and I tell my team that all the time it’s like such a blessing to be Robin Hood you know to you know

Brian Bell (00:41:44.917):

I love that so you’ve written a little bit about some writers you admire what’s the sharpest piece of writing in venture in the last I don’t know year or three six

Jeffrey Becker (00:41:57.200):

months yeah I mean we talked about Dan G a lot I mean I think that he’s just kind of come onto the scene for me in the last year year and a half I you know everyone’s following you know Peter Walker is doing amazing work with the data Carta I wrote a post recently about the people really doing the work so if you want to check out Monday morning meeting on Substack you can kind of see who I’m reading and staying up with I think Constantine from Sequoia has really sharp takes it’s really I think a unique point of view his agent swarm’s thesis was pretty interesting to me I think you know obviously Beezer on the LP side I think she’s really unearthed what it means to be an emerging manager raising and what LPs are looking for David G or sorry David Clark but also I mean back to Dan G I think his writing is some of the most like thoughtful and well researched for this inception stage I don’t think that a lot of people I just appreciate how deep he’s gone and how you know objected he’s being about the whole system so yeah I would highly recommend

Brian Bell (00:42:53.477):

people check that out what’s the best piece of advice you ever received oh you

Jeffrey Becker (00:42:57.641):

mentioned this is rapid fire and I’ve just been giving you long answers

Brian Bell (00:43:00.335):

No, you can give me a login. It’s rapid-ish fire is what I normally call it. Yeah. Wrap up questions.

Jeffrey Becker (00:43:05.438):

Honestly, I don’t know if this is advice from someone or something. I just have been evolving to believe. But we said in the beginning, to be better than average, you have to be different. There is just no other way around that. And I truly believe that if you obsess enough about things, you need to pick the things that you believe should be different and you need to exploit them. and I don’t think enough people like look at the whole picture and try to find the exploitations I don’t think they try to be different like I don’t have a lot of founders who change their pitch intentionally to stand out I don’t have a lot of people that like change the motion and the way they do things in an effort to stand out I think that most people believe a good story you know showing up and doing things that other people do it is enough and it’s just not it’s just it’s like it’s a it’s boring and B it’s like uninspiring but it’s also just not enough and I think you got to be better than if you want any better than average you got to be different and I think you got to be different in order to stand out and you got to stand out to raise capital and to attract people so that’s the thing I’m always challenging people to really think about and do and

Brian Bell (00:44:03.835):

just lean into can you think of a founder that and you’ve talked about some really great ones but you know maybe came in pretty cold cold DM cold email that just

Jeffrey Becker (00:44:12.225):

really stood out cold DM cold DM cold email I’m trying to think about this one I’m sure there are I’m sure there’s plenty I’m just trying to think of like that moment the Shosh the one I mentioned is like a just a magical human being we did a we did a further class where you can go listen to him he talks about like you know doing his investor calls at 3-4 in the morning and like you know trying to shoot arrows blindfolded and Arjun who’s like a famous mythology character but anyways the origin story of that one is I had this intern who worked for me we both went to Emory and he asked me for a job and I was like I don’t have a role but if you want to you know do an internship or come spend a few weeks here and do some stuff no problem and this guy showed up once and never saw him again never really heard from him again I figured it was just like you know maybe he had it paid off or somewhere and then a few months later he calls me and he’s like hey I’ve got this kid Shosh he also went to Emory and you got to meet him and I was like I don’t know man I don’t even you know like we didn’t even really interact I’m not sure how good this would be and so I took the phone call with Josh just kind of cold and easily in 15 minutes I was just like man I don’t know what you’re doing and what you think you’re going to build but if you want to come to New York you are more than welcome to be part of this we just started two weeks ago so you’re late but you’re more than welcome and Josh quit his job that day he was sleeping on a friend’s air mattress like basically mapping out apartments getting close to the problem and he quit his job and he flew to New York the next day he was in the office you know within 48 hours and the first day we get in the meeting we just had a session and we sort of like jammed on what we’re gonna be able to do together and I come in the next day like 7 8 a.m. he’s still there just hadn’t left just still working and it just went from like kind of cold like I don’t know what this is gonna be to you’re not really part of the cohort you’re showing up late to like wow like this kid is just gonna run circles and yeah he also just raised his seed from Canaan Partners they let Instacart Steve as well he’s got a great board and great cap table he’s just building something incredible the doorstep AI and yeah I just felt lucky to be kind of like in the ethos of that intern he thought of me to make that introduction so sometimes the best ones are just come from places you don’t expect yeah I love

Brian Bell (00:46:13.241):

that speaking of evaluating founders what’s the most overrated metric and pre-seed

Jeffrey Becker (00:46:18.343):

venture right now I mean customers revenue it’s like business changes like Harper is a great example this company I just mentioned that Raise, you know, 37 million Series A. We invested. It was a different name. It wasn’t called Harper. It was a different business. It was not in insurance at all. But if you had spent 12 seconds with Dakota, you just know that the guy is a maniac. And Tushar too. And they had built something previously. They kind of knew what needed to be done to be good builders. And so we both to check on the belief in them. Even though candidly, like I didn’t love the idea. I thought he was going up against really, really hard competitors on the previous company idea. But he, to his credit, learned that very quickly. and pivoted made a right-hand turn and pivoted into insurance and it’s been a rocket ever since and so I think like if I was only basing it on the traction at pre-seed or only on the idea I would have just totally missed it you know and now it’s you know one of the best companies in our portfolio yeah sometimes you’re yeah

Brian Bell (00:47:11.514):

I mean traction and momentum are important but so is everything else right and so everything’s a unique crystalline structure every startup’s a beautiful snowflake and you have to like evaluate that beautiful snowflake for its traction is such a

Jeffrey Becker (00:47:24.142):

noisy thing like you’re a VC and you want to be like one stage later than me you want to be like a pre-seeder seed instead of inception and you have five customers what do you do you got to call the customer as a VC your job is to get information so you call the customer one of them loves it one of them is not so sure one of them hates it now I’m confused right I’m like oh I loved it but now I’m kind of thinking not so sure then you call some people that are not customers do you need this do you like it well now you’re selling a pre-seed product it’s like what are the odds that that’s true then you kind of go a little deeper there there probably is a problem there maybe the product needs to change a bit you know it’s just it’s just the further you dig the more conflicting signals you get right and humans are emotional decision makers if you know we use the limbic part of our brains to like figure out if we like this thing or not and so you gotta like You gotta remove the noise and that’s why I like being early like if you can remove the customers and remove that stuff no one’s in your head no one’s like telling you that there’s no traction there’s traction you’re just trying to find out can this person sell or does this person know what good looks like or can this person build product or can this person move fast and so those are pretty like obvious and legible when you’re working with someone but as soon as there’s like numbers and customers you know you start to wonder is because the price goes up you’re starting to wonder is this really worth 10 15 20 million dollars and that’s a really low probability game when you look at the math of entry capital and so I try to just stay away from it I try to get the belief in the founder and that I think is less noisy and you know in some ways opens you up to these opportunities like Harper where the founder is the thing that

Brian Bell (00:48:52.690):

matters more than I love that what’s a belief that you held strongly that you changed your mind on

Jeffrey Becker (00:48:57.542):

yeah this like low price high diversification thing is like a mountain that I’ve been on you know talking about for a long time and I think there are creative ways to do deals that allow maybe a higher entry valuation with similar and sound economics for a venture fund just gotta be willing to get creative and work with founders and We’re trying to come up with ways to be a bit more flexible so we’re not missing out. You don’t want the error of omission, right? You don’t want to miss the one.

Brian Bell (00:49:22.937):

I want to say we passed on Google because we couldn’t get $500K into a $5 million cap or whatever. We couldn’t get our 10% into Google. And then you’re like, dude, you missed Google.

Jeffrey Becker (00:49:32.587):

I mean, SpaceX was certainly overvalued at $27 million after the third rocket blew up, right? That’s pretty much certainly overvalued for a company that couldn’t get off the launch pad. And nobody wanted to write that check, but the people that did are looking like geniuses right now, like David Sachs, DBL. These are storied people that actually saw it in New Milan and believed in the founder as opposed to what had happened on the launch pad.

Brian Bell (00:49:59.384):

Right.

Jeffrey Becker (00:50:00.520):

Yeah, I think that I was on this high horse about low prices and high diversification, but I’m evolving a bit on that and just coming up with new creative structures for deals that are good for our LPs that can return a lot of money, but don’t cause us to miss out on great people.

Brian Bell (00:50:15.247):

Yeah. Yeah. Rules are meant to be broken. You have to be flexible, you know?

Jeffrey Becker (00:50:18.528):

Yeah. Yeah.

Brian Bell (00:50:20.169):

Last question. What do you want your legacy to be?

Jeffrey Becker (00:50:22.670):

You know, I want to tell you a story. I don’t know what my legacy will be, but my managing partner here, who’s one that left sort of company, he’s got this idea of the cave walls, which I just love. It’s this idea that, you know, you’re born in a cave and you leave that cave and you collect firewood and you come back and light it up. And then you leave the cave again and, you know, you make memories and stories and you bring those things back and you adorn the walls with those, you know, carvings of those paintings of those memories of those thoughts. And then at some point, you know, you get married, you have kids, you know, do all these things in life. But at the end of the day, it’s still a single player game. You’re on your deathbed by yourself and you’re back in that cave and the eyes are closing. It’s like, what do you want on your cave walls? What do you want to be there? And I think about that way more often than I’m sure he thinks I do since he told me that story. I think about it a lot. I don’t know what I want the legacy to be, but I think through that lens a lot when I make decisions about, do I want to invest in this company? Will I be proud of this thing in 15 or 20 years? Where do I want to spend my time? What things do I want to do? You know, so I think the question that people should ask themselves is like, does this go on the cave walls or not?

Brian Bell (00:51:26.036):

I love that. And I think you touched on something that I think I see in a lot of, I’ve interviewed, I don’t know, probably 100 VCs on this podcast at this point and met, you know, probably hundreds more. One of the things I noticed about really good ones is the metacognition that you just described, which is thinking about how to think. How do I think about things? What’s my purpose? Would I be proud of this? It’s a lot of existential questions almost. Who do I want to be in five to ten years? You’re just constantly thinking about that stuff. I don’t know if like maybe that’s like maybe VC attracts people like that like the good like the good ones right I’ve just noticed that a lot and I’ve been I’ve been that kind of thinker for my whole life I’m just like I just think about that stuff when I worked on Wall Street 20 years ago I was like if I had 10 or 20 million in the bank would I want to keep doing this you know yeah no I don’t think I would okay like what who am I you know like there’s a lot of like medic like

Jeffrey Becker (00:52:21.773):

I think a lot of us are thinking that way right like is money the goal here right because it’s going to take us 15-20 years is this the reason I get up I don’t think

Brian Bell (00:52:31.917):

it is the goal right you look at like all the great VCs I meet they’re just like passionate and living their best lives you know and they’re not worried about like

Jeffrey Becker (00:52:41.402):

can I fly private or not who cares like yeah so you’ve interviewed quite a lot of people I’m curious to put you back on the hot seat what’s some of the best advice you’ve gotten on the show

Brian Bell (00:52:49.969):

I can’t think of anything off the top of my head you know it’s a lot of those kind of lessons though right like if I think about it a lot one of the books that I read was Seven Habits of Highly Effective People you know begin with the end in mind you know sharpen sharpen the saw you know all the all those like little lessons those little life lessons so I spent a lot of my 20s like doing a lot of that you know Tony Robbins kind of self-help stuff and trying to figure out like who I am and like what do I believe in and who do I want to be and so I’m always curious to ask people similar questions right and I just pick up it’s it’s nothing profound it’s just always like little like things like that that you sort of accumulate along the way that add up to more than the the sum of their parts kind of thing yeah I love that yeah well had a really fun time talking with you Jeff thanks so much for coming on where can folks find you online

Jeffrey Becker (00:53:37.911):

Thanks Brian I appreciate you having me here and asking questions and taking an interest in what we’re doing and also helping amplify what we’re doing I think there should be more people backing great people that’s like you know I think why we’re here and I think it’s a worthy endeavor to create real economic opportunity and change and all those things that we did at LinkedIn and that we’re doing all the way through Antler you can find me on LinkedIn obviously and on mondaymorning.substack.com is the blog you mentioned But yeah, honestly, reach out if you’re building a great company or know someone who is and would love to chat with the best, craziest, most maniacal.

Brian Bell (00:54:09.882):

Chat with the maniacs. All right. Thanks, Jeff.

Jeffrey Becker (00:54:11.947):

Thanks, Brian.

Jonathan David Nelson thinks that model is broken.

Not slightly inefficient. Broken.

Subscribe now

In this episode of the Ignite podcast, Jonathan joins Brian Bell to unpack why the startup financing machine no longer works for most growth-stage companies, why the U.S. public markets have become hostile to smaller public companies, and why the London Stock Exchange may offer a smarter path for founders stuck between venture capital and private equity.

Jonathan’s background makes him an unusual voice in capital markets. He grew up in Latin America, trained as an ER and ICU nurse, went back to school for software engineering, built Hackers and Founders into a global startup community, advised on crowdfunding policy, worked across emerging startup ecosystems, and now runs HF Capital—an AI-native investment bank focused on IPOs, secondaries, and M&A.

That mix gives him a rare lens: part operator, part hacker, part capital markets obsessive, part outsider who never agreed to pretend the system made sense.

From Trauma Nurse to Capital Markets Contrarian

Jonathan’s path into venture did not start at Goldman Sachs, Stanford, or a Sand Hill Road fund.

It started in Honduras.

He grew up in Latin America as the child of missionaries, in a small village hours down a dirt road. As a kid, he was already programming and managing his father’s mailing list. When he later came to the U.S., he expected to follow a missionary path. His father pushed him to get a practical trade first, so Jonathan became a nurse.

He worked as an ER trauma nurse before eventually injuring his back and moving into software engineering. That transition led him to Silicon Valley, where he became obsessed with startups.

His wife, tired of hearing him talk about startups nonstop, pushed him to get out of the house one night a month. That became Hackers and Founders, a meetup that started as a casual bar gathering and grew into one of the largest founder communities in the world.

But the more founders Jonathan met, the more he saw the same pain point repeat: raising capital was brutally inefficient.

Founders wanted to know where the “money tree” was in Silicon Valley. Jonathan’s answer was blunt: there is no money tree. Fundraising is a grind. It takes months. It is a brute force algorithm.

That realization became the foundation for his current work.

Fundraising Is Still a Brute Force Algorithm

One of the clearest themes in the episode is just how inefficient fundraising remains.

Jonathan compares fundraising to a brute force algorithm: knock on doors, get meetings, pitch, get rejected, and hope that one out of every ten or fifteen conversations converts.

Brian adds his own experience from raising venture funds, describing thousands of “no’s” even with an existing track record.

The point is not just that fundraising is emotionally hard. It is structurally wasteful.

Founders spend months selling stock instead of selling product. VCs spend years raising funds. LPs sit behind layers of intermediaries. Capital moves slowly through a system that is supposedly designed to fund innovation.

Jonathan’s frustration comes from seeing the full chain.

As a former nurse, he thinks in systems. In medicine, understanding how blood flows through the body tells you where to apply pressure when something goes wrong. He applies the same logic to capital markets: how does capital actually flow through the startup ecosystem?

When he realized that pension capital could pass through fund-of-funds, venture funds, and multiple fee layers before reaching entrepreneurs, he saw the system as an inefficient capital delivery mechanism.

His conclusion: the ecosystem is sick, and someone needs to heal it.

Why Crowdfunding Did Not Fully Democratize Startup Capital

Jonathan also reflects on his work around the JOBS Act and equity crowdfunding.

Crowdfunding was supposed to democratize startup investing. In some cases, it worked—especially in real estate. Real estate can generate dividends, rent, and recurring distributions. Investors have a clearer path to getting money back.

Startup equity is different.

If a small business or startup raises money from its community, when do those investors get liquidity? Usually, only through an acquisition or IPO. And those paths are not equally available to everyone.

Jonathan points out that much of tech M&A is highly network-driven. It depends on who knows corporate development teams at major acquirers. If you are outside the club, your odds of finding liquidity are much lower.

That is the core flaw: crowdfunding can help people buy private equity, but it does not solve the exit problem.

More access to illiquid assets is not the same thing as democratization.

Why Jonathan Thinks the U.S. IPO Market Is Broken

Jonathan’s most provocative argument is that the U.S. public market system no longer works for most companies below massive scale.

In his view, the U.S. exchanges are optimized for large hedge funds, high-frequency traders, and mega-cap companies. They are not well optimized for capital formation for smaller growth companies.

If a company goes public in the U.S. at a $100 million, $300 million, or even $1 billion valuation, Jonathan argues it can quickly become an orphaned public company.

The risks include:

• Limited or no analyst coverage

• High volatility

• Activist hedge funds

• Short selling pressure

• Expensive legal and compliance obligations

• Difficulty competing for attention against companies like OpenAI, SpaceX, Anthropic, or other dominant tech names

For a smaller public company, being technically public does not guarantee liquidity, coverage, or investor interest.

It can mean higher costs, more scrutiny, and less control—without the full benefits of being public.

The London Stock Exchange as a Startup Financing Hack

Jonathan’s alternative is not “never go public.”

It is: consider going public somewhere else.

He became interested in the London Stock Exchange after learning that its market structure is very different from the U.S. system. After studying dozens of global exchanges, he came away believing London has one of the best-engineered IPO products for smaller growth companies.

The appeal, according to Jonathan, includes:

• Lower IPO costs compared with the U.S.

• Lower ongoing public company maintenance costs

• A sponsor bank model

• More reliable analyst coverage

• A market maker relationship

• A less litigious environment

• Different rules and norms around short selling

• Better support for smaller public companies

The key insight is that a company doing $50 million in revenue and growing 50% year over year may no longer be a fit for venture capital—but it may be a very interesting public company in the right market.

That is the gap Jonathan wants HF Capital to serve.

The 50 and 50 Company

Jonathan describes his target company as being in the “50 and 50” range: roughly $50 million in annual revenue and growing around 50% year over year.

That kind of company can be awkward for venture.

It may not be growing fast enough for top-tier late-stage VC. It may not want private equity control. It may not want to take punishing liquidation preferences. But it may still be a strong, valuable, growing business.

In Jonathan’s view, companies like this should have more financing options.

An IPO on the London Stock Exchange could let them raise capital, create liquidity, and use public stock as a strategic asset without being forced into another painful private round.

The founder tradeoff is real. Once public, the company no longer controls its valuation in the same way. The market sets the price. Macro shocks, sector sentiment, and public investor perception can all move the stock.

But Jonathan argues that private markets have their own version of the same risk. Founders just understand those risks better because they are familiar.

The unfamiliar option is not necessarily the worse option.

Why Late-Stage Private Rounds Can Hurt Founders and Early Investors

The episode also gets into one of the least understood parts of startup finance: liquidation preferences.

When a late-stage investor puts money into a company, they may receive preferential rights that determine who gets paid first in an exit. A 2x liquidation preference means that investor gets twice their money back before common shareholders or junior preferred investors receive proceeds.

If the preference is participating, the investor may get their preference and then also participate in the remaining proceeds.

That can be excellent for the late-stage investor. It can be brutal for founders, employees, and early backers.

Brian and Jonathan discuss how late-stage investors may prefer an acquisition because liquidation preferences can matter in an M&A outcome. In an IPO, the cap table typically converts to common stock, which can wipe out those special preferences.

That creates a boardroom conflict.

Founders may benefit from going public. Early investors may benefit. Employees may benefit. But late-stage investors with structured terms may prefer a private equity acquisition.

That is why Jonathan’s idea can face resistance from boards, even when founders are interested.

Why SPACs Went Wrong

Jonathan is also skeptical of SPACs.

A SPAC, or special purpose acquisition company, is a shell company that goes public with the goal of acquiring an operating company later. In theory, it offers companies an alternative path to public markets.

In practice, Jonathan argues, SPACs are often stacked against the company being acquired.

Investors in the original SPAC can have incentives to redeem or sell, and the operating company can end up public without the same preparation, reporting maturity, or investor base it would have built through a traditional IPO process.

The result can be a brutal drop in market value after the transaction.

For Jonathan, SPACs were a financial engineering hack that often went badly wrong. His point is not that all financial engineering is bad. It is that the structure matters, the incentives matter, and founders need to understand who benefits.

Secondaries, SPVs, and the Anthropic Problem

The conversation also moves into startup secondaries.

A primary investment is when investors buy shares from the company, and the money goes into the business. A secondary transaction is when existing shareholders sell shares to other investors.

Jonathan notes that the secondary market has grown dramatically and is now a major part of venture capital.

But he is deeply skeptical of many hot-company secondary offers, especially those involving companies like Anthropic.

The issue is structure.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:

00:01 - Introduction to Jonathan David Nelson and HF Capital

01:23 - From Missionary Kid in Latin America to Trauma Nurse

03:10 - How Hackers and Founders Started as a Bar Meetup

04:26 - Why Fundraising Is a Brute Force Algorithm

05:37 - Understanding Capital Flow Like Blood Flow

08:15 - Advising the SEC and the Limits of Crowdfunding

09:40 - Why Startup Exits Remain the Broken Piece

10:47 - Why U.S. Public Markets Fail Smaller Companies

13:02 - The Origin of HF Capital and Tokenized Stock

15:26 - Discovering the London Stock Exchange Alternative

17:05 - Lower IPO Costs, Sponsor Banks, and Less Litigation

19:23 - Why Founders Still Default to U.S. Markets

21:19 - The “50 and 50” Growth-Stage Startup Profile

24:28 - When an IPO May Not Be the Right Move

26:34 - SPACs Explained and Why They Often Collapse

30:32 - Private Rounds vs. IPOs for Growth-Stage Companies

31:37 - Liquidation Preferences and Founder Dilution

35:24 - Why Boards Resist Alternative IPO Paths

36:23 - Capital Markets as a “Capital API”

38:02 - Building an AI-Native Investment Bank

40:14 - Why HF Capital Is Becoming the Bank, Not Just Selling Software

41:11 - The Coming Explosion of Smaller AI-Native Startups

42:26 - Secondaries, Latin America, and Undervalued Growth Companies

44:39 - What Startup Secondaries Actually Are

45:30 - Anthropic Hype, SPVs, and Risky Secondary Deals

47:13 - Custody, Forward Contracts, and Secondary Market Due Diligence

Transcript

Jonathan David Nelson (00:00:00.031):

Like we’re all product people. We’re all engineers, designers, you know, operations, people inside of startups. It’s no longer sexy to have like a former investment banker as part of your fund. So VCs have forgotten like how to do financial engineering. We’ve largely forgotten how to do exits. Everybody’s like, you got some of that liquidity, man. Like I would love it. Can you buy a secondary from me, man? I heard that’s the thing. You can buy my stock from me. And so like Silicon Valley is kind of stuck. We’re all just kind of trapped in this liquidity thing. It’s taken 14, 15 years now for a company to IPO. So my best performing companies in my venture portfolio, my investors have to wait 15 years for that. Like that’s a lot of crabby investors for a long time.

Brian Bell (00:01:03.042):

hey everyone welcome back to the ignite podcast today we’re thrilled to have Jonathan Nelson on the mic he spent the first part of his career as an ER and ICU nurse before stumbling into Silicon Valley that is an odd path building hackers and founders which grew from a bar meetup for or programmers into one of the largest founder communities on earth, spanning dozens of countries. Over 15 years in and around venture, he’s advised SCC on the JOBS Act and crowdfunding roles, worked with the White House and immigration as economic growth and contributed to studies on building tech ecosystems across LATAM. Today he runs HF Capital, which he describes as an AI native investment bank focused on IPOs, secondaries, and M&A. He’s one of the more contrarian voices in the market right now. He thinks the U.S. public markets are broken. interesting for everyone short of a Decacorn this will be a great conversation he’s evangelizing the London and European exchanges as a real alternative to Series B and he’s been publicly torching the wave of sketchy anthropic secondary vehicles flooding people’s inboxes he’s blunt he’s funny and he’ll disagree with me on air which is exactly why he’s here Jonathan

Jonathan David Nelson (00:02:10.061):

No, I won’t.

Brian Bell (00:02:10.642):

You’ll be here, Brian. Love it. Yeah. I’d love to start with your origin story. What’s your background?

Jonathan David Nelson (00:02:15.309):

I am a strange kid. I grew up in Latin America. Parents were missionaries. So I was the only white kid at the end of six hours of dirt road in a tiny little village at the end of Honduras called Minas de Oro Honduras I was programming managing dad’s mailing list of a hundred people that supported kind of what they were doing when I was like seven years old way back in the early 80s came to the states and was saying I was going to be a missionary dad said yeah get a trade first I was like great computer science love me some computers dad was like you know I don’t want you playing video games the rest of your life son you should be a nurse like your mom was a nurse it was very helpful when we were in the jungle I was like, okay, I was fresh off the boat. So I studied being a nurse, did that for a while, dropped out of seminary, grew a ponytail and worked as an ER trauma nurse for a number of years. I threw my back out and eventually injured out, went back to school for software engineering and I heard that you could sell ones and zeros and I know how to copy and paste so I was like I can build an app sell it for a dollar copy and paste five million times I’ll make five million dollars so easy where do I sign up

Brian Bell (00:03:23.848):

can I invest this work Yeah, I want to invest in that.

Jonathan David Nelson (00:03:26.690):

Yeah, hell. So I moved to the valley and I was working a couple miles away from Google’s HQ while I was finishing up a software engineering degree online and drove my wife crazy. She kicked me out of the house one night a month to go to this hackers and founders meetup that she helped me organize and things went viral from there. So yeah, weird backstory.

Brian Bell (00:03:46.621):

That’s so random. And then at some point you you started hackers and founders. What was the impetus to do that?

Jonathan David Nelson (00:03:53.091):

Driving my wife crazy talking about startups all the time. So I’d be like, oh, my gosh, programming language, this and nerdy things that. And she was like, honey, I love you. But yeah, You gotta get out of the house one night a month for the love of God you gotta let me help you build your business network honey so we started this meetup called hackers and founders in 2008 and after the economy collapsed a bunch of people started showing up at the meetup like In 2009 we were having a couple hundred people show up at the meetup like it was my bar night dude and we started having it twice a month and then once a week in SF San Jose Mountain View East Bay just kind of all around and people started peppering me with, they’d come up to me and say, hey, I’m here for two weeks on my visa and I need to raise $5 million. Where is the money tree in Silicon Valley? I’m like, money tree? They’re like, yeah, you go, you shake the money tree, you get a bunch of money in two weeks and then you go home. I’m like, yeah, dude, that’s not gonna happen. it’s gonna take you nine months and it’s hard it’s a brutal brute force kind of

Brian Bell (00:04:54.980):

thing fundraising is a chore like every every fund I’ve had I’m on fund three I’m closing it out probably by time this episode airs like this month in June yeah but no I mean like it’s two years of grind every single fund every single every single time and I have good track record it’s just and the same with founders it’s just a grind I mean just to close around six to nine months right

Jonathan David Nelson (00:05:14.563):

it’s a brute force algorithm it’s just you got to knock on doors you got to get the meeting you got to do the pitch you got to get turned down and you know every 10 or 15 pitches one might actually convert so if I need 50 limited partners in my fund

Brian Bell (00:05:28.494):

better find 500 limited partners to pitch basically exactly I remember fund one or fund two was my first real fund fund one was like a rolling fund fund two is my first traditional fund I was raising when you know the sky was falling in late 22 so half of my commits fell out and I think I counted up the no’s on that fund they were in the thousands for sure but like of the people that knew me knew like from team ignite the syndicate had invested with me I probably had 1400 no’s yeah 1400 yeah that’s a lot

Jonathan David Nelson (00:06:01.468):

I mean, it’s just, it’s so inefficient. And that has just always driven me crazy. And literally after the first four to 5,000 conversations about what a pain in the ass it is to raise capital, I’m like, dude, I’m the last person that people should be asking, like, how to raise capital. Like, I save lives and wipe ass. And I usually wipe a lot more ass than I do saving lives for a living.

Brian Bell (00:06:22.883):

And that was like, why are you asking? And I’m all out of lives to save.

Jonathan David Nelson (00:06:25.665):

Exactly. And I was like, how in the world? Like, why? How does this work? And the weird thing about growing up in Central America is I grew up in a currency crisis in Costa Rica. Like, the exchange rate went from 8 to 1 to 160 to 1. I’m like eight years and so you just think about money and like how does it work and what’s the exchange rate and how does it you know and just as a kid growing up in that you just kind of fundamentally think about money differently and so I just started saying you know what how does money work in in Silicon Valley. Like, it was very helpful for me as a trauma nurse to know how blood flowed throughout the human body. If someone gets stabbed and you’re leaking, where do I put pressure? You know, here or here? So how does capital flow throughout the ecosystem of Silicon Valley? And I was completely surprised when I found out that my nurse’s pension got invested into entrepreneurs through a fund of funds, through a venture fund. And when I learned that the fund of funds took 1% times 10 years, 10%, and the fund venture fund to 2% times 10 years 20% so it was 70 cents on the dollar of my nurses pension that actually got deployed yeah yeah and I’m like this is inefficient like kind of silly and like one the way back like I sell my company how does that work well the venture fund takes a 20% profit share and the fund of funds takes a 50% and then 65 cents on the dollar ends up in the hands of the nurse I’m like well this is an inefficient capital delivery mechanism man like we should engineer a new system and you know should do this and that and the other thing and you know the engineer inside of me just started saying you know the system is broken must fix the nurse inside of me saying you know the ecosystem is sick must heal and that kind of led me on this long torturous journey and now I’m starting an investment bank man because this process this process shall not stand man

Brian Bell (00:08:15.824):

And I love this and I want to get to HF Capital but there’s a couple more things I want to talk about because you’ve done a couple of really interesting things. You actually advised the SEC as I mentioned in the intro on the JOBS Act on Title III crowdfunding almost a little over 10 years ago. Yep. This was supposed to democratize startup capital. Did it work and or did it just produce a lot more noise in the system?

Jonathan David Nelson (00:08:35.551):

It worked for a certain category of companies. It works really well for real estate because real estate offers dividends, you know, right?

Brian Bell (00:08:44.255):

There’s some distributions coming in every quarter and yeah.

Jonathan David Nelson (00:08:47.037):

And so Investors have a way of actually getting some money back and it’s pretty easy to fractionalize that you know I get a million bucks in rent on a annual basis and I have a thousand investors everybody gets you know a thousand bucks a year and so that’s actually The problem is, is that how if they buy stock in a crowdfunding, when do they get to resell that stock at a profit? Well, the two ways that you do that are through an M&A and through an IPO. 90% of the world’s tech M&A happens within 60 miles of my front door here in San Jose, California. And it’s all very clubby. Who do I know who knows someone at Google corporate development who might actually be interested in buying my engineering team because I’m out of money. And so if I’m a small business owner in Minneapolis and I want to build, I don’t know, a barbershop, I need to fund money, raise money to actually grow my barbershop. I should be able to crowdfund for my customers. But how are they ever going to be able to sell that stock? They’re not realistically. So that’s like the bug in the system is this like exit piece. And our stock market in the United States is optimized for massive hedge funds and high frequency traders. And so they are not optimized for the SEC’s third mandate, which is capital formation. And they pretty much only work for like SpaceX and Anthropic. and massive multi-billion dollar valuation companies if you like go in the United States at a hundred million dollar 300 million dollar valuation like you’re kind of

Brian Bell (00:10:22.258):

screwed yeah why is that why why are you kind of screwed there a couple of reasons

Jonathan David Nelson (00:10:27.682):

with a small cap IPO in the United States you run a couple of risks one is if you get a hedge fund that’s an activist they buy five million bucks of your stock or five percent of the company they can and demand a board seat. Then they can start arguing for you to do stuff. Hedge funds can short your stock. And in the United States, you don’t actually have to borrow the stock. It’s kind of a trust me, bro, short system. Like I’m going to short the stock. And then they can tweet about it because you can talk to your own book and if a large hedge fund starts shorting $100 million stock as an entrepreneur like I’m screwed and so you get these massive volatility kind of swings how do I if I’m a $500 million if I’m a billion dollar company in the United States on the US stock markets I’m like number 4,000 on the S&P 5000 on the Russell 5000 index right right how do I a billion dollar valuation how do I compete with all the noise and all of the media and all of the earned media that SpaceX is getting or Anthropic is getting like I can open AI which is kind of yeah or you know top 20

Brian Bell (00:11:33.670):

tech Cerebris who just IPO recently yeah how the hell do I compete with them I

Jonathan David Nelson (00:11:37.872):

don’t so how do I convince an investment banking analyst to cover my stock and then they market and sell that analyst report to like institutional investors it’s just not going to happen So below $5 billion in market cap in the United States, the vast majority of those companies are just orphaned. They get no analyst coverage. $5 billion is what a medium-sized hedge fund. So there’s all of these risks that are kind of built into the United States stock market system. It’s just too big and it’s been too successful.

Brian Bell (00:12:11.605):

And then so, you know, fast forward now to HF Capital, the AI Native Investment Bank that you’re starting. What was the origin story there?

Jonathan David Nelson (00:12:18.756):

So I was thinking we could tokenize stock, you know, if I could build a global digital stock certificate that could could trade on blockchain with, you know, could it be legal in 185 different legal jurisdictions around the globe and not get me thrown in jail? Like you can do that, but you have to IPO or do the reporting in compliance to actually be able to let retail investors, you know, the poorest 97% of the country to actually buy that stock legally per the SEC’s rules. So I was going to do that, but yet you need tools to streamline and automate that process. Publicly traded quote unquote venture fund that would have tokens instead of limited partner shares. I got the SEC, the police inside of the SEC to verbally sign off. I was like, dude, will you write something so I can show it to like investors? They’re like, yeah, no, we don’t do that. And so rich people in Silicon Valley had no idea why I would ever want to do this. Like, why is it needed? We don’t need it. We don’t need it. Like there’s tons of exits. People outside of Silicon Valley didn’t believe A in blockchain and B that I could ever do this without going to jail. So it struggled. Somebody at the London Stock Exchange. I bumped into Chris Mayo at the London Stock Exchange and they started talking about how their IPO are vastly different. And I was like, no. And it’s common in the UK if I’m doing 10, 20, $30 million a year in revenue, like I can IPO in a London exchange. And all of my concerns are like, well, what about hedge funds? Well, it’s illegal to short stock like you guys do in the United States and the UK. You actually have to borrow the stock from someone and sell it. And the people that invest in these small cap IPOs are like long only. I’m like, okay, well, how do I get analyst coverage? Well, you get a sponsor bank, you get a chaperone, and you’re guaranteed analyst coverage. I’m like, well, what about volatility? I’m like, well, you know, no. You have, like, well, how do I, if I have to trade 10% of my stock, like one of my investors wants out, how do we do that? I said, well, because you get a sponsor bank, you get a market maker, and they will pick up the phone and make some calls for you. And you might not sell it immediately, but you can sell it over the next three to six months without tanking the stock. I was like, how the hell have I never heard about this? Like, how do you screw me? They’re like, well, we’re the London Stock Exchange, dude, we can’t really screw you. So I was insanely curious about this, very suspicious. And so being the engineer, I need to know how the system works. So I’ve talked to 28 stock exchanges now looking at different listing regimes around the globe. And by God, they have the best, like the best engineered IPO product. They struggle marketing it because you’re a 400 year old brand and why would we need to market you know the London Stock Exchange and there’s just kind of some legacy stuff there the listing seems fantastic but it’s you know it is what it is so you scout you scoured the market

Brian Bell (00:15:08.580):

for all the different stock exchanges and the best the best one especially for it sounds like small cap mid cap that you found was London Stock Exchange maybe talk a little bit about some of the things they do you talked about the sponsored bank which is interesting So you always have this market maker and analyst coverage you literally have to put up collateral on a short which probably radically reduces the actual shorting of a stock right because you got to put up some money to do that. What else did you find with the

Jonathan David Nelson (00:15:41.302):

We will have more shareholder lawsuits in the next six months in the United States than London has had in the last 10 years. Because in the UK, the loser has to pay for the lawsuit, has to pay the legal costs for the other side, much less litigious. In the United States, the SEC says, hey, if we’re going to be public, you have to fill out all of these forms. And the lawyers who are really good at filling out those forms charge you about two grand an hour. Like the S1 and the S1 registration statement, K forms, the Q forms, all of that stuff. If I need to, well, if I want to raise more money, I need to file either an S1 or an S3. Each one of these S1 forms, it’s like a 1,000, 1,500 hours of legal time at 2,000 bucks an hour, you’re looking at like one to 3 million bucks, crazy expensive. And so just maintaining a listing in the United States, you’re looking at anywhere between one to $3 million. To IPO, you’re looking at anywhere between five to like 60 million bucks. in the UK you get a sponsor bank you hire a regulator like it’s a licensed physician it’s called a nominated advisor they do most of the most of the form filling out for you they make sure that you’re doing things okay their lawyers cost about 20% of what our institution what our public markets lawyers do here their accountants on this lower end of the market they have a bunch of boutique accountants boutique law firms that do this and they’re much cheaper so all in to IPO instead of five to 60 million bucks, you’re talking about one to four. Your annual maintenance costs are probably about $350,000, $450,000. And it’s a completely different kind of regulatory mindset. And venture capital is kind of a latecomer to the UK. And so companies just got less used to raising money on the London Stock Exchange. Like that’s how companies in the UK and Europe have grown for 400 years.

Brian Bell (00:17:37.512):

So why don’t more companies do this? What’s the friction there where everybody wants to IPO on the NASDAQ or New York Stock Exchange versus the LSE? U.S. has much better marketing, much better financial press.

Jonathan David Nelson (00:17:51.292):

You know, like when was the last The U.S. Capital Markets Press doesn’t cover them. Founder we have 20 years worth of venture capitalists in Silicon Valley probably more like 25 that don’t understand capital markets like we’re all product people we’re all engineers designers you know operations people inside of startups it’s no longer sexy to have like a former or investment banker as part of your fund. So VCs have forgotten like how to do financial engineering. We’ve largely forgotten how to do exits. Like everybody’s like, you got some of that liquidity, man. Like I would love, can you buy a secondary from me, man?

Brian Bell (00:18:37.125):

I heard that’s the thing you can buy my stock from me.

Jonathan David Nelson (00:18:39.646):

And so like Silicon Valley is kind of stuck. We’re all just kind of trapped in this liquidity thing. It’s taken 14, 15 years now for a company to IPO. So my best performing companies in my venture portfolio, my investors have to wait 15 years for that. Like, that’s a lot of crabby investors for a long time. And so, but we don’t know anything else, you know, and here this punk guy who, you know, has got a ponytail and goes around saying, you know, trust me, bro, the London Stock Exchange has a better system. And the people in town are like, are you crazy? Are you a communist? Like, are you well? And I’m like, I’m a hacker, dude. Like, this is a hack.

Brian Bell (00:19:16.523):

at what point would a startup want to consider IPOing versus raising a series I

Jonathan David Nelson (00:19:21.769):

don’t know DEF whatever I’m looking for companies in like the 50 and 50 range 50 million bucks a year in revenue 50% year over year growth like growing 50% year over year close to profitable that is not a venture fundable company Right. Like Sequoia will not write you a check.

Brian Bell (00:19:37.650):

No, if you just went from 25 to 50 mil, or let’s call it, I guess that would be 35 million of ARR to 50, that’d be roughly 50% growth. And you’re roughly EBITDA, you know, cash flow positive. You’re just not venture backable anymore at that scale. I mean, you might be able to find a VC firm to write you a check, but

Jonathan David Nelson (00:19:57.915):

Yeah You give me a tech stock that’s growing profitably 50% year over year I will yellow that bitch every day of the month Like, I’m in Like, are you kidding me? And I get audited financials And I get quarterly reporting And, you know, financial transparency Yeah, like these should be public stocks. They should not be private companies. But the private equity industry, the venture capital industry in this, everybody in the United States is like, no, no, no, no, don’t IPO. It’s a nightmare. Don’t IPO. It’s a nightmare. It’s a nightmare. But I think it’s a great hack, dude. The benefits to the founder are, it’s a lot less. My due diligence essentially is done in my IPO process in London. It’s done. I report financially, like everything about me is transparent. So if I want to raise a follow-on round, I talk to my sponsor bank, say, hey, dude, I want to acquire this company or I want to open up this, you know, other market, blah, blah, blah. Do you think I could actually raise another 50 million bucks? Your sponsor bank will say, let me make some calls and they will come back and say, yeah, we can You can do the deal. Like two, three days later, a week tops, you close your check. And as a founder, I didn’t have to do that fundraising. Like my investment bank, they’re an outsource sales team. So when I heard this, I’m like, wait, wait, wait, wait. I can raise follow on rounds in days? If I want to acquire a company, I can just issue more stock. I can just acquire another company in stock. I can start rolling up other companies. I’m like, why isn’t everybody doing this? How do you screw me? I think it’s just it’s a marketing thing. People just don’t understand it. They’re terrified of it. Boards are terrified of it. VCs who are on the boards are like, I don’t want to screw up the best company in my portfolio. No, let’s try to find a private equity fund we can sell you to.

Brian Bell (00:21:45.843):

That’s interesting. So why would a company not do this? So kind of take the steel man the other side. So, hey, I’m a founder. Maybe I’m growing faster than the 50%, right? So I’m more venture backable. That would be one reason we kind of covered that. But are there any other reasons why you wouldn’t consider

Jonathan David Nelson (00:22:04.039):

I no longer control my stock price. So as a founder, when I want to sell my stock, I go out and fundraise. And if I no longer say this is my valuation and I don’t sell, if they want to sell at a much lower price like I have that choice as a publicly traded company you don’t get that liberty and so something happens to the stock market a war happens in the Middle East my share price drops AI starts to happen I’m a SaaS company people think oh my gosh I’m gonna get killed you know my share price drops like there are things out of your control that affect your share price and that sucks The other side of that is I never have unhappy investors because if my investors are unhappy, they sell my stock. If they believe in me, they own my stock. So it’s a different flip. It’s a different switch that you have to flip. You can’t talk as much about the internet. you should probably hire an investor relations and PR firm like that should become you know you no longer just talk to VCs and talk to you know build your company you’re still talking to public investors but you’re generally talking to them through the press so that’s kind of a different flip if I’m nerdy that’s hard right And honestly, I did work with a company, took them to IPO. Zay, after a lot of work, the board decided to launch their own SPAC in the United States. And a SPAC is like a clean company that doesn’t have any operations.

Brian Bell (00:23:39.470):

Explain what SPAC means. It sounds like an acronym for people who don’t know what that is.

Jonathan David Nelson (00:23:43.042):

It’s a total acronym. So it’s a special purpose acquisition company. The only purpose of this company is to one IPO. It’s just it’s a paper. It’s an on paper company. It’s a shell company. You convince an investment bank to let you take it public. You convince investors to put money in it. And then my job as the owner of the SPAC is I go around and I look for another company to acquire. It’s a special purpose acquisition company. So when I see a company that I want to acquire, I acquire them with stock. The SPAC investors can either sell the stock, sell their own stock in the SPAC, or they can roll it over into the new company. There are some quirks with how we do SPACs in the United States, that it’s very easy to SPAC at a $300 million valuation. All of the investors in the original SPAC ghost. And now I’m trading at a $2.5 million valuation. And that’s what happened with the company that I worked with because the board decided to launch their own SPAC because it was sexier. It’s 23, 24. Everybody’s doing this. We know how to manage the risks, blah, blah, blah. I’m like, dude, it’s a bad idea. It’s a bad idea. It’s a bad idea. You should just IPO in London instead. They put a bunch of money into the SPAC,

Brian Bell (00:24:58.397):

but they couldn’t take that money in stock fast enough to acquire the company, the target company. So it lost all its value.

Jonathan David Nelson (00:25:05.121):

The SPAC had 60 million bucks in it. Investors had put money into the SPAC. They get warrants. So if the SPAC acquisition goes well, they can buy more stock at a discount. If it doesn’t go well, if they don’t like the acquisition, They can just sell. And so there’s all of these incentives for people to actually buy into the shell and to withdraw their money after the acquisition is complete. So that happens all the time.

Brian Bell (00:25:31.416):

This is why I hear in the news all these SPACs collapsing in value so fast.

Jonathan David Nelson (00:25:35.739):

So what happens is the SPAC does the acquisition. The other company becomes public through this reverse merger process. It’s financial engineering. All of a sudden, all the investors are like, Bail! Everybody sell! And the share price ends up just tanking. It’s pretty normal for a SPAC to drop 85, 90% of its value in the first year.

Brian Bell (00:25:58.376):

Wow and is that because they don’t have the same lockup restrictions as regular IPOs or?

Jonathan David Nelson (00:26:02.559):

It’s that it’s kind of a way around SCC rules because the company that IPO is doesn’t have any operations so their filings are very cheap when they acquire another company now the company has to build all of that they haven’t done this before they have to build this internal muscle they have to figure out how to do it they’re probably not good at public reporting yet and The investors had an incentive, a big financial incentive to have the acquisition and then sell their stock. And so it’s SPACs are stacked against the company that gets acquired.

Brian Bell (00:26:35.871):

So let’s say I’m a startup and I have 50 million of revenue. Yes. So basically I can go out, probably can’t raise 50 million on, I don’t know, call it 250, right? Because I’m only growing 50% a year. Maybe I can raise, I don’t know, 10 on 50 million to revenue growing 50%. What do you think that valuation would be in venture? It’s probably 150, probably 3x of revenue, something like that. So I can go raise maybe 15 on 150. It’s all 10% of the company, maybe 30 on 150. What would be kind of the terms on the London Stock Exchange IPO process that you think you could get? So are they better typically?

Jonathan David Nelson (00:27:10.968):

If you are raising privately at that stage, I am probably selling what’s called liquidation preferences. That investor is probably getting a lot of insurance.

Brian Bell (00:27:21.131):

Yeah, which kind of messes up people like me earlier on the cap table right now, because now they have 2x LICPREF. Maybe you can explain what that is for people listening and haven’t heard that before.

Jonathan David Nelson (00:27:30.533):

So again, it’s another bit of financial engineering. A liquidation preference is when a company gets liquidated or exit, I and preferred I get a preference and how much money do I get back before anybody else gets that money back so if I have a 2x liquidation preference I invest 10 million bucks into this company when this company IPOs the first 20 2 million dollars in liquidity comes back to me because I have a liquid liquidity.

Brian Bell (00:27:57.827):

I have a 2x liquid pref. Yeah.

Jonathan David Nelson (00:27:59.508):

And so I get 2x my money back and everybody else who’s lower did a 1x liquidation preference because we’re founder friendly and we want to be in an earlier stage.

Brian Bell (00:28:09.856):

And the company was growing 5x back then when we invested.

Jonathan David Nelson (00:28:13.198):

Exactly. And so the early investors end up getting screwed. The founders end up getting screwed.

Brian Bell (00:28:18.471):

Right, let’s say the IPO for, I don’t know, let’s say it’s a, or they exit, they get acquired by a PE firm for $100 million. So the first $20 million of proceeds is going back to that late stage 2X Lickpref investor. Now there’s $80 million left over. for all everybody else yep but the it depends if it’s participating or

Jonathan David Nelson (00:28:38.716):

non-participating maybe explain that as well so participating preferred is I get 2x and then I get 10% of the rest like I bought 10% of the company the first 2x I come and then the participation is I participate the same as everybody else does

Brian Bell (00:28:55.178):

Now I still get 10% of the 80 million.

Jonathan David Nelson (00:28:57.401):

So I get another 8 million. Wow. So I get $28 million. For my $10 million investment. On my $10 million investment. It’s a great investment.

Brian Bell (00:29:05.931):

Right. That’s a little 3X. And if you just invested two, three, four years ago, your LPs are pretty happy.

Jonathan David Nelson (00:29:10.517):

Absolutely. And so from like an asset managed perspective, it’s brilliant. It’s not founder friendly. Founders are almost always very optimistic about their companies. And if they’re up against the wall, though, they got to do what they got to do to keep the company going. And so they they will take bad terms if I IPO that company everything my whole cap table converts to preferred stock converts to common stock so I sell common stock that 2x liquidation preference VC or PE firm probably bought a board seat as well and they’re going to start putting the thumbscrews to me to do whatever is best they’d

Brian Bell (00:29:46.675):

rather see the acquisition But if you IPO, my understanding is all the Lickpref gets crammed down to common. So you bought 10% of the company, you wrote a $10 million check at $100 million, you’re just going to get 10% of that IPO, whatever that is. Versus like an acquisition. Now the acquisition cascades down the Pref stack. That’s why the VC is the late stage VCs will steer a company towards an acquisition. They’d rather see their buddy down at the PE firm buy the company because now they’re going to stand to 3X instead of 2X.

Jonathan David Nelson (00:30:17.413):

And chances are the guy, the buddy at the PE company is probably a limited partner in my fund. I mean, that’s just the game. You can’t hit the players. Don’t hit the players, hit the game.

Brian Bell (00:30:26.134):

Yep. Hit the players, hit the game. Exactly. But here comes Jonathan with an idea, right? To the founders. And so the founders bought in. This is this sounds great. I’m going to get more money as a founder, right? All my employees and myself and my my co-founders. And now I got to go sell this to the board. And that’s therein lies the friction right there.

Jonathan David Nelson (00:30:46.478):

Yes. And board members are like, why have I never heard about this? Same reaction I had. How do you screw me? Like what’s the downside? No, like SPACs were creative. That was a hack. And that went really, really wrong. And, you know, there are risks. You know, what if I can’t sell my stock? What if it’s a down market? What if my industry isn’t sexy anymore? You still have the same risks of that in the private market, but you understand how to.

Brian Bell (00:31:11.769):

You know what you’re getting into? There’s a little bit of like a jumping off the cliff with the IPO, right? Yeah. Because now I’m bringing it out to the market.

Jonathan David Nelson (00:31:17.371):

It’s fundamentally changing how your company runs.

Brian Bell (00:31:19.412):

Right yeah and then there’s a cost right what does it cost to do another another raise privately at that late stage let’s call it the 50 million dollar revenue company growing 50% a year and I’m going to go raise another 10 to 20 million versus IPO is there a cost savings there?

Jonathan David Nelson (00:31:36.462):

Privately depends on how much you value your time. If it’s in a private market, and if you’re taking a flat round or a down round, you’re probably looking for those investors for 12 to 18 months and your company’s growth is going to stall, you know, blah, blah, blah. It’s the time. value I then find an investor I do diligence I am chasing I’m selling my stock instead of selling my product that is what it is if I IPO I do diligence for once and for all and then I maintain it and it’s I as an engineer inside of me I think of capital markets stock markets as a capital API as long as I comply with these rules these are my API keys I can plug into the capital market who have been created to get capital to the companies that need it when they need it.

Brian Bell (00:32:24.473):

This is my swagger doc for the... Yeah.

Jonathan David Nelson (00:32:28.076):

And so engineers kind of understand the API in the capital markets, but I’m probably the only person that talks about it that way. And so it’s been a hard sell, frankly. I probably pitched three I had a handful kind of bite and the boards stopped them all so next step is I’m probably going to raise a single purpose growth fund where I invest in the company I take a board seat on the condition that that I walk the company through the IPO

Brian Bell (00:33:02.499):

process oh that’s interesting now you’re on the board and you can kind of have a little bit more sway

Jonathan David Nelson (00:33:07.207):

And I have skin in the game. And the founder sees it not as Jonathan saying, trust me, bro, you should totally IPO here. It’ll be awesome. Oh, and I, by the way, you’ll have to pay me for the privilege. Two, I’m buying your stock. I’m putting my capital at risk. and we will do this together. I’m on your board. So I can’t really run away after the IPO. Like I got to be with you. So that’s the direction that I’m heading. But this entire process, like, dude, there’s probably only 10 or 15 things Ignite Insights Ignite Insights You have access to all the prospectuses of every public company for the last 40 years. You can train an AI on that. You can generate it and have a lawyer just supervise it. Double check it. Human in the middle. All my compliance. Like that should be AI. have a human in the middle to be sure you’re not hallucinating you should be able to have an AI map the universe of public investors and IPOs because that’s public so we should be able to know who to talk to for what kind of company and when there’s all sorts of things that can be automated and I’m like why isn’t anybody else doing this I out of those 10 or 15,000 people I’m probably one of like a handful of product guys so that’s why I’m building my AI native investment bank

Brian Bell (00:34:35.132):

Fascinating. So why not build, I mean, I’m sure you thought about this. Why not build the tools and just sell them to the investment bank versus become the investment bank yourself?

Jonathan David Nelson (00:34:44.819):

Sure. You know, it’s SaaS. I can sell you a SaaS product and that’s great. I get monthly recurring revenue or I can sell you a SaaS product and I can use it myself and I can charge a commission. SaaS plus.

Brian Bell (00:34:59.118):

And I think I think your model is interesting because I think what’s happening and I see it in the early stage because we invest precedency right we do some late stage secondary stuff which we can get into but for the most part it’s precedency right and I see more startups being created than ever getting more traction than ever

Jonathan David Nelson (00:35:16.283):

Yes.

Brian Bell (00:35:16.543):

But there’s also this there’s higher highs like the Anthropics and OpenAIs and etc. But I think there’s also this like fatter tail of outcomes.

Jonathan David Nelson (00:35:24.632):

Yes.

Brian Bell (00:35:25.113):

That are going to need more options. Right.

Jonathan David Nelson (00:35:28.396):

Absolutely. And it’s getting there fewer and fewer companies IPOing. It’s taking a lot longer. I think Having been at Hackers and Founders for years in 2008, 2009, I was probably the tip of the spear. And I was like, y’all need to be ready. There’s going to be an order of magnitude more founders created now that I have Stripe and now that I have access to open source and AWS and cloud.

Brian Bell (00:35:50.936):

Yeah.

Jonathan David Nelson (00:35:52.857):

AI there’s going to be another order of magnitude or two more startups they’re going to be smaller by definition and ventures kind of scaling but it’s probably four or five years behind investment banking is 15 years behind the ball on this like nobody that I’ve come across well there’s actually a YC company that’s focusing on AI and M&A they have like an M&A An engineer and an M&A investment banker kind of working side by side. Fantastic. But like there’s Silicon Valley doesn’t understand this. This is like a New York thing. It’s like a London thing.

Brian Bell (00:36:27.948):

yeah so you’ve also through practical VC worked on secondary side so these are LPGP interests and and funds tell us more about that so after the last IPO that I worked

Jonathan David Nelson (00:36:39.651):

on 2002 22 23 I got a little burnt needed to take a break friend of mine Dave McClure said hey I’m building a secondaries fund could you help me buy some secondaries in Latin America you’re from there you’ve invested down there for years I’m like yeah sure what do you want it’s like I want one investment like okay when six months he ended up liking the discounts and the quality of this the quality of the companies that he was buying into that I think it ended up being like 15% of his funds number two just because he is a value growth buyer so he looks for the best value in the company with the strongest financials and in Latin America these companies like there’s two funds that raise that do growth investing down there and if they pass you’re kind of screwed and so these companies just sell and sell and sell and they have to be profitable and so This company’s doing two, $300 million a year in revenue, growing 80% year over year with a 10% profit margin, dude. And where are they going to IPO? Mexico? No. United States? Not for another five years. So we were one of the only, there’s a couple more now, but we were the only buyers of secondaries in the region and it’s a great business great fund I mean tiny tiny LP in one of his funds and I’m like thrilled it’s gonna do great and so but I wanted to kind of throw I brought my own shingle into the iBanking thing and so that’s kind of what I’m doing but secondaries market is now as large as the primary market the secondaries do people in your audience understand what a secondary is is that you can explain

Brian Bell (00:38:17.841):

it for people who might be listening to understand what that is yeah

Jonathan David Nelson (00:38:21.003):

so it’s a New York financial engineering term so the primary investment is when the company itself sells stock to investors and the money that the investors invest go into the company that’s the primary investment any kind of secondary sale is when investors sell that stock between themselves investors sell to each other and so you’re buying second hand stock so it’s a secondary sale and so right now there is a very quickly growing market for buying and selling secondaries, i.e. everybody wants to buy a little bit of Anthropic. Oh, my gosh. Oh, my gosh. Oh, my God. Do you have any Anthropic?

Brian Bell (00:39:00.244):

Do you have any? Oh, man. I get emailed every day. Exactly.

Jonathan David Nelson (00:39:03.763):

I get like five people a week asking me if I have access to Anthropic because I’m in Silicon Valley and I’m like yeah you don’t want that because it’s going to be

Brian Bell (00:39:10.366):

scammy by the time it’s offered to you I just saw one today it was 1.3 trillion so that’s like literally almost 50% over the latest valuation and was it through a

Jonathan David Nelson (00:39:20.490):

multi-layer SPV yep they all are and so what happened how that works is I can actually buy like 50 million bucks of Anthropic stock from one of the entrepreneurs. Well, I can’t buy the stock. I can write a contract to buy the stock in the future at a price.

Brian Bell (00:39:36.916):

That’s a forward contract. That’s different.

Jonathan David Nelson (00:39:38.496):

But it’s a forward contract. I put that contract into a special purpose vehicle and SPV, which is just a shell company that holds a contract. Jonathan David Nelson (00:39:46): and then I can sell pieces of that SPV to other investors who are like, ooh, I’m going to create my own SPV out of this. And so I’m going to buy chunks of this SPV and then I’m going to create my own SPV and I’m going to go and find other investors so it’s like this group of Russian nested dolls where you have an SPV that owns a part of an SPV that owns a part of an SPV that owns a forward contract on stock options and if you actually look at where those SPVs are incorporated you know you might catch one in Panama you might catch one in the Cayman Islands a lot of these tend to be

Brian Bell (00:40:23.502):

sketchy yeah you gotta you gotta be careful and we’ve done a lot of these at team ignite and we’re always making sure there’s there’s a custody of like shares and the due diligence because you can end up in these yeah forward contracts or like yeah it’s like multiple shells through all kinds of Caribbean islands yeah so yeah

Jonathan David Nelson (00:40:40.757):

you gotta be careful there yep absolutely but it’s the secondaries market is as big as the primary market in venture capital right now

Brian Bell (00:40:48.265):

it’s crazy but a lot of times the companies themselves have rofer right so one of

Jonathan David Nelson (00:40:55.212):

the couple of challenges one is according to SEC rules I can’t have it see the 2000 or 5000 shareholders before I actually have to report like a publicly traded company before I have to fill out that awful S1 registration statement so as a private company I need to control who’s owning my shares and if I have a bunch of people selling little pieces parts of my company all over the place like that’s a regulatory danger for me and if they’re scammers selling this stuff or shady or or various shades of gray. I’m taking the reputational hit on the value of my stock. I’m no longer controlling the value of the stock. There’s a market kind of, I’m raising money at a $900 billion valuation. Eanthropic someone offers you to sell the stock at a $1.4 trillion valuation 1.3 on one hand that’s nice because then I can actually go back to my primary investors and say oh look on the secondary market my stock is 1.3 you’re getting it from easily 900 billion but the company wants to control it so they have these rights of first refusal written into their incorporation whereas they get to buy the stock first before anybody else actually gets to buy it before you sell it to anybody else So you usually need to have the board to buy in to selling the secondary if it’s actually secondary shares. Right. You generally need board approval for that.

Brian Bell (00:42:20.507):

That’s what a lot of people don’t realize is if you are going to buy the shares on the secondary market and actually transfer the shares, the company needs to approve. And most likely a hot company like Anthropic will not approve. They’ll say, no, we’re going to buy those shares.

Jonathan David Nelson (00:42:32.657):

Correct.

Brian Bell (00:42:33.578):

Yeah. Let’s talk about the future like what are you excited about over the next you know five or ten years as you kind of look at this these categories developing this IPO market hopefully taking off your AI driven investment bank taking off. What are you excited about?

Jonathan David Nelson (00:42:48.958):

I mean, I’m excited about the tools that I’m building internally. I have been a genetically engineering as opposed to vibe coding. I have been vibe coding ton of internal tools, which have been like truly giving me like superpowers. Like I had a massive network before. because I ran hackers and founders. Like the tools that I’m actually building with AI and putting that in AI and connected to other different APIs, I feel like I have like Jarvis for fundraising right now. And I’m going to have Jarvis for IPOs and I’m going to have like Jarvis M&A as well. I will probably have Jarvis for secondaries. And so that has been mind blowing as to how much I’ve been able to build in a short amount of time and how much leverage I think that’s going to give me in the future.

Brian Bell (00:43:35.760):

Right, right.

Jonathan David Nelson (00:43:59.410):

the future of blockchain I have said for years is building new capital markets infrastructure payments infrastructure stock settlement real estate transactions loans bonds all of that is going to be on chain at some point in time and it’s probably going to be emerging markets that are starting to leapfrog the United States in some of these technologies yeah So like Venezuela, Argentina, you know, the economies are terrible. They, Venezuela, half of the country’s economy runs on like crypto.

Brian Bell (00:44:30.316):

Right, on stable coins and, right.

Jonathan David Nelson (00:44:32.457):

It’s not a micropayment. It’s just how you pay for things. East Africa, almost all of Kenya’s economy is built on a currency that’s based on cell phone minutes. Wow. So you can, it’s called M-Pesa and Vodafone accidentally crashed. I created it because you could transfer minutes to each other and you can sell the minutes and people just started using minutes to pay for things and it’s now like East Africa and PESA is like the euro for large stocks of East Africa. and sell Vodafone cell phone minutes and pay for things like on feature phones. Like that in my mind is leapfrogging what’s happening in the United States. I’m much more excited about what’s happening in emerging markets than I am in the US. You know, I love to be some Stanford You know, I grew up in Latin America, helping Latin Americans have access to the same capital markets tools as the gringos have in the United States or East Africa, like the

Brian Bell (00:45:31.232):

Why hasn’t like a Mark Zuckerberg level founder just said, nope, if you want to buy into my company, it’s like, here’s the token. Why hasn’t anybody enforced that or has any well-known unicorn founders done that yet or done that? DecoCorn founders said nope we’re completely blockchain if you want to buy our company here it is here’s the token it’s freely tradable you can buy in anytime you

Jonathan David Nelson (00:45:52.794):

can sell anytime why has anybody done that in the ICO craze was probably 2017 so we’re only nine years out there is a crypto company and I’m spacing on the name they’re actually going to be IPOing it like they’re doing like three four hundred million dollars a year in value in valuation they’re Basically their equity has been the tokens that they issue, but it’s different than their stock. And so, you know, they haven’t really needed to sell stock because they sold tokens along with their equity at the same time. They will be public probably in the next 24 months. they’re an Asian company a lot of crypto companies have issued you know Binance issued the Binance coin all of these crypto companies haven’t needed they’ve sold a product which has been their blockchain tokens as opposed to selling their stock and so they haven’t really needed to sell their stock but yes bonds need to be on the blockchain there’s so many more efficiencies if you can get the blockchain to actually operate fast enough to actually all of these assets are essentially going to be on chain at some point it’ll take another 10 years

Brian Bell (00:46:54.102):

It’s amazing. I learned so much about the IPO process. Where can the founders and VCs and anybody interested get in touch with you?

Jonathan David Nelson (00:47:02.664):

Yeah, no, absolutely. Email me, jay at hf.capital. And I’m, you know, pretty active on LinkedIn. So Jonathan Nelson at LinkedIn. Happy to talk. Happy to chat. And if you give me a chance to geek out about capital markets, I’m in.

Brian Bell (00:47:16.886):

Yeah, well, it was a lot of fun. I learned a ton. Thanks so much for coming on.

Jonathan David Nelson (00:47:19.767):

Really appreciate it, Brian. Thank you.

That sentence has a funny way of compressing a week into a calendar invite. Over the past several week, we met dozens of YC Spring 2026 companies. Some calls were polished. Some were messy. A few felt like watching someone open a door before the rest of the room knew there was a door there.

It was a very strong batch.

But the thing I kept noticing was not just company quality. It was the market around the companies. Rounds were moving, but they did not seem to close with quite the same violent speed as some recent YC batches. Founders were still getting term sheets. The best companies still had heat. But the whole thing felt a bit less like a stampede.

Part of that may be boring, which means it is probably true. It is June. A lot of investors are on vacation. We usually are too. Venture capital loves to pretend it is immune to calendars, but it is still run by humans with families, flights, and school schedules.

The other part was price.

The new center of gravity seemed to be around a $30 million valuation cap. That felt like the standard ask for many companies. If a company had real traction, it was not unusual to see $40 million or $50 million. Those numbers are no longer reserved for the obvious monsters.

That changes the job. At a lower price, you can make some mistakes if the team is excellent and the market is moving. At $40 million or $50 million, the company has to clear a much higher bar. You are no longer asking, “Could this become interesting?” You are asking, “Is this already showing signs of becoming a category winner?”

That is a different question.

And it made this batch especially interesting, because the batch had a very clear shape.

The simple summary is that YC Spring 2026 was an enterprise batch. Out of 196 companies, 120 were B2B. If you add industrials, fintech, and healthcare, you get roughly 91% of the batch. Consumer was present, but thin. This was not a batch full of new social apps, travel products, or consumer marketplaces. It was a batch full of companies trying to make software do work that humans currently do.

That sounds obvious until you sit through enough pitches and hear the same pattern from different angles.

A founder shows an agent that can run customer operations. Another automates credentialing for healthcare. Another builds an AI-native ERP for manufacturers. Another gives agents phone numbers. Another creates authorization for agents. Another builds a browser layer, a memory layer, a runtime layer, an observability layer, or a control layer.

At some point you stop hearing “AI startup” and start hearing something more specific.

Software is becoming labor.

That was the real theme of the batch. The stronger companies were not selling better dashboards. They were selling work getting done. A claim gets processed. A shipment gets scheduled. A compliance task gets completed. A QA test gets maintained. A medical billing workflow gets handled. A manufacturing process gets monitored. A government records request moves forward.

This is a big shift.

The last generation of software mostly helped people do their jobs. It organized information, reduced coordination costs, and made teams more productive. The new generation is trying to become part of the workforce. It does not just sit next to the employee. It takes over a loop.

That is why so many YC companies sounded like “AI employee for X” even when they used more sophisticated language. The phrasing varied, but the bet was consistent. If an AI system can own a narrow workflow from start to finish, the buyer stops comparing it to software and starts comparing it to headcount, outsourced labor, error rates, cycle time, and revenue leakage.

That is where budgets live.

It also explains why regulated and operationally ugly categories were so prominent. Healthcare, insurance, compliance, government workflows, manufacturing, logistics, and defense all showed up with real density. These are not glamorous markets from the outside. They are full of forms, approvals, messy edge cases, old systems, and people doing repetitive work because the cost of being wrong is high.

That ugliness is the opportunity.

A lot of great startups begin in places where the workflow looks too annoying for a tourist. If a market is easy to understand in five minutes, it is usually crowded in five months. The best enterprise markets often make you want to quit the diligence halfway through. Acronyms pile up. The buyer is hard to identify. The sales process has strange rituals. The product has to integrate with software that looks like it was designed during the Bush administration.

Then you realize the customer pays a lot because the pain is real.

That is why I found the healthcare and fintech parts of the batch more interesting than the raw counts suggest. In healthcare, the serious companies were not just chasing consumer wellness. They were going after credentialing, contracting, FDA regulatory work, billing, labs, clinical operations, imaging governance, and specialty practices. In fintech, there was a notable cluster around insurance, asset management, market infrastructure, and prediction markets.

Insurance in particular kept coming up. That makes sense. Insurance is a giant bundle of risk assessment, distribution, regulation, paperwork, pricing, and claims. AI can touch many of those pieces, but the hard part is accountability. If you can underwrite, distribute, service, or administer insurance in a way that is faster and more accurate, you are in budget territory fast.

The same logic applies to healthcare administration. Nobody wakes up excited to buy medical billing software. That is the point. The buyer does not need to be entertained. They need fewer denials, faster collections, cleaner workflows, and fewer humans stuck in repetitive administrative loops.

There was also a real hard-tech signal in the batch. Industrials was the second-largest top-level category, with companies in manufacturing, robotics, defense, energy, aviation, space, and drones.

This was easy to miss because AI software dominated the language. But underneath that language, a meaningful number of founders were building for the physical world. Counter-drone systems. Submarine drones. Compact nuclear reactors. Space manufacturing. Data center cooling. Robotic arms. Factory technician tools. Welding intelligence. Industrial sourcing. Manufacturing ERP.

A lazy read would be, “YC is just funding more AI SaaS.”

That misses the more interesting thing. YC seems to be funding AI as a control surface for the real economy. The software is increasingly connected to machines, factories, clinics, logistics networks, labs, government offices, and defense systems.

That is a healthier signal than another batch full of generic chat products.

The crowding is still real. Very real.

Agent infrastructure was everywhere. Runtimes, browsers, memory, monitoring, sandboxes, workspaces, authorization, phones, coding tools, company brains. Some of these companies will be important. Many will get absorbed, bundled, outpaced, or flattened by the platforms beneath them.

The danger is confusing a layer with a company.

A layer can be useful without becoming a durable business. A tool can be impressive without controlling the customer relationship. A demo can be magical while the product sits one platform update away from irrelevance.

This is where the batch became harder to evaluate. In a normal software market, you can underwrite a startup around product quality, founder quality, customer urgency, and distribution. In an AI market, you also have to ask whether the company is building on ground that will still exist in twelve months.

That does not mean “avoid infrastructure.” Some infrastructure companies will win huge. But they need a reason to persist. They need to become a control point, not a temporary patch. They need usage data, distribution, deep integration, trust, compliance, switching costs, or a technical advantage that improves as the market gets more capable.

The same is true for “company brain” products. There were several versions of the idea that every company needs a central memory layer for AI employees. I believe the direction. I am less sure that every version becomes a venture-scale company. The question is who owns the system of record when AI becomes part of the org chart.

That question matters more than the demo.

The consumer side was almost the inverse. Only 12 companies were categorized as consumer. Even there, many were still AI-shaped: personal agents, AI-native gaming, creator tools, consumer simulations, AI companions, and interfaces for spawning agents.

This says something about the current funding environment. Changing consumer behavior from scratch is brutally hard right now. Enterprise buyers have pain, budgets, and workflows. Consumers have infinite apps and limited patience.

That does not mean consumer is dead. It may mean the next great consumer company will look strange at first. The best consumer products often begin as toys, status objects, or habits that serious people dismiss. But this batch was clearly tilted away from that risk profile.

YC seems to be saying that the near-term AI opportunity is inside work.

I mostly agree.

The highest-conviction companies in this batch were the ones that had three things at once: a painful workflow, a buyer with budget, and a path to owning more of the operating loop over time.

That last part matters most.

It is easy to automate a task. It is harder to own a workflow. It is much harder to become the place where the work lives. The best companies in this batch were trying to climb that ladder. The weaker ones felt like clever automations that might get copied, bundled, or ignored once the novelty wears off.

Experienced investors spend a lot of time trying to separate those two. The question is rarely “does this work?” anymore. Most demos work well enough. The better question is, “If this works, who cares, who pays, and what does the company get to own next?”

That is where the batch split.

Some companies felt like sharp tools. Others felt like the beginning of operating systems for a narrow slice of the economy. In venture, that distinction is everything. Tools can make money. Operating systems can become massive.

The best way I can describe YC Spring 2026 is this: the batch felt like a map of where founders think AI will first win budget authority.

Not attention. Not novelty. Budget authority.

That is a useful lens. AI has already won attention. It has already won demos. It has already won the right to be tried. The next phase is less forgiving. Companies need to show that they can save money, make money, reduce risk, or own work that someone already pays humans to do.

That is why the batch was strong even though parts of it were crowded. The direction of travel is right. The easy wrappers are getting squeezed. The serious companies are moving into painful workflows where the product has to survive contact with reality.

Prices are higher. Rounds may be a little slower in June. Investors may be on boats, planes, or pretending not to check email from vacation.

But the underlying startup formation is not slowing down.

The question after this batch is not whether AI will create important companies. That part feels settled.

The question is where the durable value lands.

My bet is that the winners will be the companies that stop sounding like software vendors as fast as possible. They will become operators, departments, infrastructure, insurers, brokers, labs, analysts, compliance teams, and factory systems. They will take responsibility for outcomes, not just generate outputs.

That is the hard version of the AI startup story.

It is also the only version worth paying YC prices for.

Two things made the moment stranger than the headline number.

SpaceX was not the only giant reaching for the door. Three days earlier, OpenAI confirmed it had quietly filed paperwork with securities regulators to prepare for a listing of its own, while taking pains to add that it had not committed to any timing and that going public carries a complicated set of tradeoffs. Two of the most valuable private companies on earth moved toward the public market in the same five days.

The second thing was quieter and matters more. While those two companies walked toward public shareholders, the money funding the rest of the AI build changed character. On June 9, Broadcom, Apollo, and Blackstone announced a financing vehicle with an initial 35 billion dollars behind it, aimed at more than a gigawatt of Anthropic’s compute and a plan for over 20 gigawatts of capacity by 2028. That is not a venture round. It is the way you finance a power plant or a toll road, with lenders who want repayment schedules and a return, not a manifesto.

Put the three together and a pattern shows up. For most of their lives, these companies answered to believers. Private investors who bought the vision, wrote large checks, and waited. This week, the people they answer to started changing. Public shareholders. Credit funds. Lenders. Strangers, in other words, and strangers do not buy vision. They price risk. The whole week was the sound of the audience swapping out.

Strangers do not buy vision

When a private company gets a public price, something happens to everyone still holding the private version. They suddenly have a real number to argue with. For years the value of a stake in SpaceX or OpenAI was settled in whisper networks and the occasional employee sale, where the price was whatever a small group agreed to that month. A listing replaces the whisper with a quote that updates every second.

That helps the strong names and exposes the weak ones. A company with real revenue and a credible path to profit now has a public comparison it can point to when it argues for a higher mark. A company whose only support was narrative adjacency, the claim that it sits near AI or near defense and therefore deserves the same multiple, just lost its hiding place. The public market is a harsh appraiser. It does not care how good the story sounded at dinner.

SpaceX itself is a useful warning against treating a hot debut as a verdict on quality. The company reported revenue near 19 billion dollars last year, up about a third, against losses in the billions. A 19 percent pop on day one is a price signal driven by scarcity and a wall of retail demand, and scarcity is not the same thing as durable economics. The right read is not that the company is worth whatever the first day said. It is that public investors now get to vote daily, and their votes are not sentimental.

Experienced investors will spend the next few weeks watching whether that public appetite holds past the opening rush, because the answer sets the tone for the listings still in the pipeline. If the strangers stay enthusiastic, the door stays open for OpenAI and the next wave. If they cool, every private mark that leaned on the assumption of an easy public exit gets harder to defend.

Robots get the infrastructure treatment

The same shift, money behaving like infrastructure finance rather than a venture bet, showed up this week in a place that used to be pure science project. Physical robotics.

On June 10, NEURA Robotics announced a Series C of up to 1.4 billion dollars to scale its cognitive robots and the platform it wants other robots to learn from. The backer list reads like an industrial atlas: Amazon, Nvidia, Qualcomm, Bosch, the European Investment Bank. The company says it is sitting on an orderbook above a billion dollars, a figure worth treating as the company’s own claim until deployments confirm it. A day later, Standard Bots said it had raised 200 million at a one-billion-dollar valuation, pitching robots that learn by being shown a task rather than being programmed line by line, and claiming it is on pace to supply a meaningful slice of new industrial robot installations in the United States next year.

Notice what the strongest pitches in robotics now emphasize. Not the dexterity demo. The orderbook, the factory capacity, the deployment pipeline, the data each installed machine sends back. The bet has moved from can the robot do the trick to can the company turn a fleet of robots into a learning system that gets better with every customer. Robotics is being priced as embodied AI infrastructure with a data loop, and capital is showing up in amounts that match that ambition.

For founders, the line between the two halves of this market is becoming the whole game. A humanoid that looks impressive in a video and has no path to installed-base learning is a demo. A boring inspection or maintenance robot that gets smarter across thousands of sites is an asset. The capital this week rewarded the second kind and will keep grading the first kind harder.

The platforms took the doormat

While the financiers rearranged the top of the stack, the largest consumer platforms spent the week absorbing the layer a lot of startups hoped to own.

At its developer conference on June 8, Apple unveiled a rebuilt assistant it calls Siri AI, with the ability to read what is on your screen, pull personal context across your messages and email and photos, take actions inside apps, and run as its own dedicated app. Around the same time, Meta pushed its Business Agent deeper into WhatsApp, Messenger, and Instagram, handling customer questions, bookings, lead qualification, and catalog sales, and wiring into systems like Shopify and Zendesk. Meta also struck a data-center deal with Reliance in India, starting at 168 megawatts with room to grow, a reminder that these companies are now planting AI capacity next to the demand rather than renting generic cloud.

When a platform that ships to a billion devices builds the horizontal assistant into the operating system, the standalone consumer assistant becomes a brutal place to start a company. Same story for the generic small-business chatbot once the social network that owns the customer’s inbox gives the feature away. What survives platform absorption is workflow ownership the platform cannot easily see. The vertical tool that lives inside a system of record, the compliance layer in a regulated industry, the data interoperability play that cleans messy operational information across systems that do not talk to each other. Poetic’s 50 million dollar Series A this week, built around deterministic execution for high-stakes processes like fraud and insurance work where a free-roaming agent is a liability, is a tell about where defensibility is moving. Reliability as the wedge, not raw capability.

The biggest platform is the government

The week’s sharpest reminder of who controls the frontier did not come from a lab or a cloud provider. It came from Washington.

Anthropic disabled its two newest and most capable models, Fable 5 and Mythos 5, to comply with a United States export-control restriction, while leaving its earlier models running. The company’s own statement is the firm part of this story. The machinery behind it, which agency and which authority, has been described through unnamed officials and is worth holding loosely until a primary document shows up. The shape of the event is not in question. The most capable thing the company had shipped went dark, fast, by order rather than by choice.

Consider what that does to a valuation. Every private mark on a frontier lab has priced talent, compute, revenue, and competition. None of them priced this. The chance that a company’s single most valuable product gets pulled overnight, for reasons that have nothing to do with demand, sat on a risk slide as a tail scenario. This week it became a line item.

For the companies built on top, the lesson is portability. A product whose only source of intelligence is one closed model, reachable through one company’s servers, now carries a risk it did not last month. That access can be cut, and not because the provider chose to cut it. So a few unglamorous categories look smarter than they did a week ago. Middleware that can fail over from one model to another in minutes. Inference you can run on your own hardware or inside a sovereign boundary. Open-weight tooling that treats model access as something to govern rather than rent. The thin wrapper on a single frontier model looks more fragile every time the ground under it moves.

This is where the export story rejoins the rest of the week. The new owners of these companies, the public shareholders and the lenders underwriting their data centers, are the same people who put this sort of risk in writing. A believer waves it off. An underwriter prices it, discloses it, and asks for a discount. A model that can be ordered offline becomes a risk factor in a public filing and a covenant question in a credit agreement. The shift that opened the week, from belief to terms, is what turns a regulatory surprise into a lasting markdown.

The open question is whether access comes back, and on what terms. Treat any rumor of a quiet reversal as rumor until the company or the government says so plainly.

The two things money cannot fix

Here is the twist that ties the week together, and it arrived from the least glamorous corner.

On June 10, the Labor Department reported that consumer prices rose 4.2 percent over the past year, the hottest reading in three years. Energy did most of the damage, responsible for more than 60 percent of the monthly increase. Strip out food and energy and the picture was calmer, with core prices up 2.9 percent, so this is an energy shock more than a broad reacceleration. The practical effect is the same either way. The assumption that cheaper money is coming soon to rescue long payback periods is off the table, with the next Fed meeting only days out and a cut nowhere in sight.

This is the constraint the financiers cannot structure their way around. You can externalize compute into a credit vehicle and turn a data center into an asset that lenders underwrite. You cannot finance away the price of electricity, and you cannot finance away the cost of borrowing when the central bank is pinned by inflation. The entire AI build runs on those two inputs. Both got more expensive or less certain this week, in the same breath that the build’s financing moved onto balance sheets that have to service debt.

So the companies converting capital into energized, permitted, cooled capacity look better than ever. The companies whose margins quietly assumed cheap, stable power and a friendly rate cut look more fragile than their decks admit. That gap between assumed inputs and actual ones is where a lot of 2024-vintage math is going to break.

What this means for the founders we back

The thread running through all of it: the people funding and holding these companies are shifting from believers to underwriters, and underwriters ask harder questions. The right response, at every stage, is to own something an underwriter can defend. A workflow, a data set, a deployment surface, a margin structure that survives a rate that stays high and a power bill that does not fall.

What looks more attractive after this week is concrete. Software that turns robot fleets into data assets. Tooling that makes AI capital expenditure legible to a lender or a public investor, the unglamorous accounting of utilization, energy cost, and model cost. Workflow agents that execute reliably in domains where an error is expensive. Energy procurement and power-aware operations, which are quietly becoming board-level concerns rather than facilities footnotes.

What looks worse is equally concrete. Horizontal consumer assistants competing with the operating system. Small-business chatbots living only inside someone else’s messaging app. AI apps with high token burn, thin workflow ownership, and no control over their model, their distribution, or their data. Humanoid robotics sold on spectacle with no deployment loop underneath. And late-stage secondary positions whose only justification is that SpaceX and OpenAI are headed public, which is borrowed confidence, not analysis.

A handful of questions worth putting to founders this week:

• If a platform gives away the horizontal assistant layer for free, which part of what you do still has value?

• How sensitive is your gross margin to token prices, energy prices, and a model provider raising its rates?

• Do you own proprietary operational data, or are you renting intelligence and distribution from someone larger?

• In anything physical, what loop makes your system measurably better after each customer install rather than just bigger?

• Could a lender or a public-market investor underwrite your infrastructure dependencies from the numbers you have today?

One more signal for the people who fund us rather than the people we fund. Kindred Ventures closed 355 million dollars in new funds this week. Capital is still flowing to specialist early-stage managers with credibility in this cycle, even with valuations running hot. The appetite for concentrated, expert access to the frontier has not cooled. It has gotten more selective about who gets to be the expert.

What we are watching

Whether SpaceX holds its enthusiasm past the opening week, because that answer sets the temperature for every listing behind it. Any public disclosure that follows OpenAI’s confidential filing, which will hand the whole private AI stack its first real comparables. More compute deals structured as project finance, using chips and capacity as collateral. Apple’s developer terms for Siri AI, since the economics of every assistant startup on the platform depend on what Apple chooses to expose. And whether this week’s robotics megadeals convert into deployed machines, or stay as orderbook claims in a press release.

The week’s real lesson is not that the money returned. It is that the money changed its mind about what it is buying. For a long time the frontier ran on belief. This week it started running on terms.

A Danish-born, New York-based serial founder, Dennis has built and sold four companies, shut one down, and is now building his sixth venture, LaunchBrightly. His career has moved through analytics, media optimization, AI scheduling, and now product documentation automation. Along the way, he has learned a hard truth that most startup advice skips over: the best companies are not always built around glamorous ideas. They are built around painful, persistent problems that someone is already paying humans to solve.

That is the thread running through this episode of Ignite.

Dennis is not interested in founder theater. He does not angel invest while building. He does not sit on boards. He does not treat advising startups as a badge of honor. His view is blunt: building a company is already improbable enough. Why make the odds worse by scattering your attention?

That philosophy comes from experience.

The “Expensive MBA” of Startup Failure

Before the exits, Dennis had a failed startup he still refers to as an expensive MBA.

He tried to build what was essentially an early Grubhub-style business in Europe, but with a twist: instead of running a pure marketplace, his company controlled the customer-facing brand and relied on third-party food operators behind the scenes. The model gave him better margins and helped revenue ramp quickly. But there was a structural problem underneath the growth.

He did not control the quality of the product.

Customers blamed his brand when food quality slipped. Partners had little loyalty. And while the business looked promising from the outside, the underlying mechanics were broken.

Dennis had a chance to merge with Just Eat, one of the major European players in food delivery. He turned it down. Four months later, his company was dead.

The lesson was not simply “take the acquisition offer.” Dennis is more precise than that. The real lesson was that the market was telling him something and he refused to listen clearly enough. He was too attached to the original model when the market was shifting toward marketplaces.

That distinction matters. Founders are told to be persistent, but persistence can become delusion when the market is obviously rejecting the mechanism, not the mission.

Dennis still dislikes dramatic pivots. He does not romanticize the classic startup story where one company morphs into something completely unrelated and becomes a massive success. To him, that is often luck dressed up as strategy. But he does believe founders must be willing to adjust the business model when the pain is real and the current solution is wrong.

In his case, the pain was real. The model was wrong.

Build From a “List of Hate,” Not a Whiteboard

One of Dennis’s strongest founder habits is deceptively simple: he keeps a “list of hate.”

Instead of sitting in a room with smart people, ordering pizza, and brainstorming startup ideas from a blank whiteboard, Dennis tracks the things that annoy him in real life. Bad workflows. Broken processes. Repeated friction. Small moments where the world feels unnecessarily stupid.

He does not immediately act on the list. He lets it accumulate over years. Then, when it is time to start something new, he studies the list and looks for recurring pain.

Some ideas get deleted because they are just personal irritation. Others have already been solved. But the valuable ones show up again and again—and those are worth investigating.

This is a powerful founder filter because real pain tends to persist. Technologies change. Interfaces change. Distribution channels change. But the underlying job-to-be-done often remains stubbornly intact.

Dennis used that process across multiple companies. It helped lead him to Visual Revenue, X.ai, and eventually LaunchBrightly.

Market Challenge or Science Challenge?

Dennis also uses a sharp framework for evaluating startup ideas: are you attacking a market challenge or a science challenge?

A science challenge is something people clearly want, but the technology is hard. Self-driving cars are the obvious example. If cars could reliably drive themselves, the market would exist. The question is whether the technology can actually work.

A market challenge is different. The technology may be straightforward, but customer behavior is uncertain. Airbnb was not hard because the website was impossible to build. It was hard because people had to become comfortable sleeping in strangers’ homes or letting strangers sleep in theirs.

Dennis’s warning is simple: know which one you are attacking. Better yet, avoid attacking both at the same time.

X.ai was a science challenge. People already hated scheduling meetings. The fantasy was obvious: someday, when they climbed high enough in the organization, they would have an assistant who handled scheduling for them. The demand was not mysterious. The question was whether software could do the job.

And in the pre-LLM era, that was brutally hard.

Building AI Agents Before AI Agents Were Cool

Long before ChatGPT made AI agents mainstream, Dennis was building X.ai, an AI scheduling assistant that could coordinate meetings over email.

Today, that sounds almost obvious. At the time, it was anything but.

X.ai had to solve scheduling through natural language before modern large language models existed. That meant building much of the machinery from scratch: annotation tools, intent classification, entity extraction, workflows, and huge labeled datasets.

Dennis says the company hand-labeled 32 million data points. They identified 47 scheduling intents, including new meetings, rescheduling, running late, adding participants, and changing durations. They focused on three core entities: time, location, and people.

This is what building AI looked like before the current tooling stack existed. Before the modern agent hype cycle, the work was painfully manual.

And Dennis learned something important: he became too attached to making the AI feel human.

He wanted to win what he calls the daily Turing test. He wanted users to believe the assistant was human. That was intellectually thrilling, but commercially distracting. If users did not know it was a machine, they did not know they could buy one for themselves. And when the assistant made mistakes, people judged it differently because they thought they were interacting with a person.

Eventually, Dennis accepted the more useful product truth: it is okay for a machine to behave like a machine.

A button can be better than a sentence. A workflow can beat a performance. The goal is not to fool the user. The goal is to solve the pain.

That lesson feels especially relevant now, when the AI market is full of products trying to look magical instead of becoming reliable.

Selling vs. Being Bought

Another major lesson from Dennis’s journey is the difference between selling and being bought.

Early in a startup’s life, founders are selling. They are convincing customers that the problem matters, that the product works, and that the company deserves a chance. But at some point, if the market is real, the dynamic changes. Customers already know they have the pain. They arrive with better arguments than the founder could give them.

At that point, the job is no longer to sell harder. The job is to make buying easier.

That sounds obvious, but many startups miss the transition. They keep optimizing the sales motion when the market has already moved into demand capture. Dennis argues that founders need to recognize when buyers are no longer being educated from scratch. Once people are trying to buy, friction becomes the enemy.

Why Dennis Talks to M&A and Corp Dev Early

Dennis also rejects a common piece of startup advice: avoid corporate development and M&A conversations because they are a waste of time.

His view is the opposite. Talk to people. Keep them updated. Build relationships long before you need anything.

Every one of his exits came from this kind of long-term relationship building. A pitch that seemed irrelevant later led to Yahoo acquiring IndexTools. A casual conversation with another startup eventually led to the introduction that became the X.ai acquisition.

The point is not to constantly shop the company. The point is to build optionality.

Founders often underestimate how much future opportunity comes from seeds planted years earlier. A 30-minute conversation can look useless in the moment and become decisive later.

LaunchBrightly and the Value of Boring Problems

Dennis’s current company, LaunchBrightly, is aimed at a problem most people would never call exciting: keeping product screenshots in help centers up to date.

But that is exactly why it is interesting.

Every software company ships product updates. Every update risks making documentation stale. Old screenshots confuse customers, create support tickets, slow down onboarding, and make the product look worse than it is. The faster engineering ships, the harder it becomes for documentation, support, and product marketing teams to keep up.

LaunchBrightly automates that process. It logs into a software product, captures updated screenshots, scans help center articles, detects mismatches, and helps teams update documentation without manual screenshot drudgery.

This is not glamorous AI. It is not a general-purpose agent promising to change the world. It is a narrow, painful workflow that already exists inside software companies.

That is the point.

Dennis likes markets where the customer already has a human doing the work. You may or may not buy his product, but you cannot avoid the problem. If your documentation is stale, you pay somewhere else: in support costs, customer confusion, and slower product adoption.

The Founder Who Still Loves Zero to One

After four exits and one failure, Dennis is still building.

Not because he needs another trophy, but because he loves the sport. He is unusually honest about this. Many founders say they love building, but what they really want is the outcome: the exit, the status, the headline, the fundraise. Dennis is drawn to the part most people hate: the zero-to-one phase where there are few customers, few believers, and a high likelihood of failure.

That is where he feels most alive.

His view is that if you are building only for the trophy, you may never get one. But if you genuinely love the game, you are more likely to survive long enough to win.

This episode is not a neat founder success story. It is more useful than that. It is a conversation about judgment: when to persist, when to adapt, when to ignore the glamorous idea, and when to chase the boring pain that refuses to go away.

Dennis Mortensen’s career is a reminder that great startups are not always born from inspiration. Sometimes they come from a list of things you hate, a refusal to die, and the discipline to solve the problem exactly as it exists.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:
00:01 – Meet Dennis Mortensen

01:25 – From IBM Dreams to Serial Founder

03:51 – Selling His First Company During the Dot-Com Era

05:00 – Building in Budapest and Moving to New Yor

07:08 – Why European Founders Look West

09:25 – The “Expensive MBA” Startup Failure

11:52 – Why Dramatic Pivots Are Overrated

13:51 – The Marketplace Mistake That Killed the Business

16:27 – When the Market Is Telling You You’re Wrong

18:23 – The Twitter Pivot and Founder Mythology

20:46 – Why Business Model Flexibility Matters

22:35 – Founder Bias, Persistence, and Not Dying

24:30 – Shutting Down and Moving On

26:34 – Building IndexTools and Real-Time Analytics

31:34 – Why Founders Should Take M&A Calls

35:05 – How Optionality Creates Future Exits

36:45 – From Yahoo to Visual Revenue

40:01 – The “List of Hate” Startup Ideation Process

44:57 – Why Founder Focus Beats Angel Investing

48:43 – Building Visual Revenue for Digital Publishers

53:44 – Selling Visual Revenue to Outbrain

54:58 – The Pain Behind X.ai

55:26 – Market Challenge vs. Science Challenge

56:59 – Why Scheduling Was a Worthy AI Problem

01:00:40 – Testing X.ai with Human Assistants First

01:02:31 – Wizard-of-Oz Testing and Scheduling Complexity

01:05:34 – Building AI Before Modern LLMs

01:06:07 – 47 Intents and 32 Million Labeled Data Points

01:10:13 – Lessons from the X.ai Journey

01:11:14 – Why Winning the Turing Test Was the Wrong Goal

01:14:55 – When Customers Stop Being Sold and Start Buying

01:17:04 – Introducing LaunchBrightly

01:17:43 – Building for the Love of the Sport

01:19:20 – Why LaunchBrightly Exists

SpaceX filed to go public, and the moment it did, the private market for its stock went dark. The company began its roadshow on June 4, the stretch of meetings where executives pitch big investors before a listing, and it is expected to set a price around June 11 and start trading the next day under the ticker SPCX, at a value reported close to 1.8 trillion dollars. Whatever that opening price turns out to be is now the price. The quiet side door I had been using closed for good. And because SpaceX absorbed Elon Musk’s AI company xAI back in February, buying SpaceX today means buying one of the largest artificial intelligence bets in the world, folded inside a rocket company.

SpaceX was not alone at the door. On June 1, Anthropic, the maker of the Claude family of AI models, confidentially submitted a draft of its own listing paperwork to securities regulators. A confidential filing lets a company start the process in private, sharing its financials with regulators before the public ever sees them. It came days after Anthropic raised roughly 65 billion dollars at a valuation approaching a trillion. OpenAI, the company behind ChatGPT, is reported to be weeks away from filing too.

So three of the most valuable privately held companies on earth, all of them soaked in AI, are heading for the same exit at once.

Here is why that matters more than any single product launch. For most of the last decade, these companies sold a story. They raised money from people who believed, and belief is patient. A public listing ends the patience. The day a company lists, it has to reconcile its story with audited numbers, every three months, in front of strangers who can sell the instant the math disappoints them. The believers get replaced by scorekeepers. That is a different sport, and the whole industry is about to learn whether the numbers underneath the narratives can survive daylight.

You could see the early grading in the private market itself. On the same trading platforms, Anthropic now changes hands at an implied value near a trillion dollars while OpenAI sits closer to 880 billion. That gap did not exist a few months ago. It opened because Anthropic’s coding tools started throwing off real revenue and because almost nobody who owns the stock wants to sell. Scarcity plus revenue moved the price. Meanwhile names like Databricks, a data and analytics company, and Stripe, the payments company, traded all week in tight, liquid ranges, which tells you investors have made up their minds about what those two are worth. The market is no longer guessing about them. It is pricing them.

The squeeze nobody put in a press release

The IPO drama is loud. The quieter story this week is the one that should worry an ordinary founder more, because it is happening to the ground they build on.

Start with a move that looks like plumbing and is closer to a land grab. On June 1, OpenAI made its top models and its coding agent generally available on Amazon Web Services, the cloud computing service that runs a huge share of the internet’s back end. Companies that already buy their computing from Amazon can now reach OpenAI through the same billing and security they already trust, with no separate contract. That sounds like a routine distribution deal until you remember who Amazon is. Amazon is Anthropic’s largest backer and its main cloud partner. OpenAI just set up shop inside its biggest rival’s house, and the landlord helped move the furniture in.

The same week, Microsoft shipped a family of its own AI models, built in-house, so it can lean less on both OpenAI and Anthropic for the intelligence inside its products. Two of the largest buyers of frontier models are quietly building their own.

Then the floor dropped out of the price. A category called open-weight models, where a developer publishes the inner workings of an AI system so anyone can download and run it, kept closing the gap with the expensive leaders. One such model from the company MiniMax now costs about twelve cents for a million units of text input where a leading closed model charges five dollars for the same volume. The unit there is a token, the small chunk of text an AI reads or writes (more or less a root word), and the bill for any AI product is mostly a count of tokens. Twelve cents against five dollars is not a discount. It is a different economy.

Put those three together. A startup whose only real advantage was access to the best model now gets pressed from above, where the cloud giants own the model and the customer, and from below, where nearly-as-good intelligence is almost free. The middle is a thin place to stand.

A public company showed you what standing there costs. On June 2, GitLab, which sells software that helps engineering teams write and ship code, reported revenue up 23 percent to 264 million dollars and announced it was cutting 14 percent of its staff in the same breath, about 350 people, while exiting 22 countries. The chief executive framed the cuts as the company reshaping itself for an era of AI agents doing more of the work. Read that again. A profitable, growing software business looked at the AI era and concluded it needed fewer humans, and it said so in an official filing rather than letting it accidently leak. Expect more growing companies to make the same trade out loud, because now one of them has shown it plays well with investors.

And the workflow layer itself is being bought up. On June 4, Cloudflare, an internet infrastructure company, acquired the small team behind Vite, a building tool that a large fraction of web developers use every day without thinking about it. When a giant absorbs a tool that millions of builders rely on, it gains a quiet say over how their work gets deployed. The value keeps migrating toward whoever owns the distribution, the daily workflow, and the data, and away from whoever simply wraps someone else’s model in a nicer interface.

Two governments and one AI lab reached for the same lever

For years the official posture toward AI was to stand back and let it run. This week, in three different places, the hand reached for a lever.

Anthropic widened a program it calls Project Glasswing, which hands a tightly restricted, unusually capable model to organizations that defend critical systems so they can hunt for security holes before attackers do. On June 2 it extended that access to about 150 more organizations across more than 15 countries, in power, water, healthcare, communications, and hardware. The company says the model has already surfaced more than 10,000 serious vulnerabilities since the spring. A day later, Anthropic published what it learned watching a year of people trying to misuse its models for attacks, and the finding that lingers is that the share of more capable bad actors grew over the year. The tools that help defenders also help the other side.

Governments noticed. On June 2 the White House issued an executive order on AI and security that sets up a process to test frontier models for cyber capability and stands up a Treasury-led clearinghouse to share what they find, while pointedly refusing to require any company to get a license before releasing a model. The government would rather measure the risk than gate it.

Across the Atlantic, Britain’s competition regulator did something sharper. On June 3 it ordered Google to let publishers keep their articles out of its AI-generated search answers without losing their normal place in search results. Until now, publishers faced a grim choice. Let Google’s AI summarize your work at the top of the page and watch your traffic fall, or vanish from Google entirely. The regulator broke that bind and called it a first of its kind, then said the harder fight, whether Google has to actually pay for the content it uses, would wait at least another year. Google said it would roll the controls out globally.

None of these three moves was coordinated. That is what makes the week notable. An AI lab and two governments independently decided in the same stretch of days that the sharper edges of this technology need tripwires and bargaining rights around them. For anyone building in security or in the economics of content, the rules just became a live variable instead of a background hum.

The money stayed expensive

While the technology accelerated, the cost of money refused to fall.

On June 5 the monthly jobs report landed well above what forecasters expected, with 172,000 jobs added against estimates near 80,000, unemployment holding at 4.3 percent, and wages still rising more than 3 percent over the year. Earlier months were revised upward too. A labor market this sturdy gives the central bank every reason to keep interest rates where they are rather than cutting them, which is what a lot of growth plans had quietly assumed would happen by now.

There is a new hand on that lever. Kevin Warsh was confirmed in May as the new chair of the Federal Reserve, and the Fed’s June 16 and 17 meeting will be his first. The president who appointed him wants lower rates. The data keeps arguing against them. That tension will shape the cost of capital for everyone for the rest of the year.

For the giant AI buildout, much of it financed with borrowed money, expensive capital is a headwind that compounds. For young companies, it brings back an old discipline. When money is cheap, burning it looks like ambition. When money is dear, spending less to get further is the thing experienced investors reward, and that pressure is heaviest at the earliest stages, where a pre-seed or seed-stage company, the first real outside money a startup raises, has the least room for error.

Where the next surprise is hiding

If you want to know where the ground will shift next, watch where the smartest money and the biggest labs are pointing at the same time. This week they pointed at robots.

A young company called Generalist AI raised 400 million dollars at a valuation of about 2 billion, backed by serious names including Jeff Bezos and the computer scientist Fei-Fei Li, to build the control software that lets robots handle delicate physical tasks. The company claims a large jump in success on fiddly manipulation, the sort of thing human hands do without thinking, though those figures come from the company itself and have not been checked by outsiders, so treat them as a claim rather than a fact. Nvidia, the chipmaker whose hardware sits under most of modern AI, pushed further into robotics with an open blueprint for humanoid machines. OpenAI started hiring for robotics. Three separate signals, one direction, in a single week.

Two of the people who think hardest about the long arc gave the week its mood. On the Moonshots podcast hosted by the entrepreneur Peter Diamandis, the inventor and futurist Ray Kurzweil held to his long-standing forecast that machines will match human intelligence by 2029 and that the deeper transformation arrives around 2045. He argued that AI already outpaces people on certain narrow mental tasks by a wide margin, and that the laggards are physical understanding and robotics, the messy business of acting in the real world. Set his comment next to where the money went this week and the point sharpens on its own. The thing he named as furthest behind is exactly the thing the labs and the investors just rushed toward. When the acknowledged weak spot becomes the hot destination, that weak spot is usually where the next surprise lives.

Underneath all of it sits a constraint that no amount of capital removes quickly. The head of the world’s most important chip manufacturer, TSMC, told shareholders on June 4 that the largest technology companies are on track to spend around 725 billion dollars on AI this year and that supply will trail demand for a long time. SoftBank committed tens of billions to new AI data centers in France. The bottleneck has moved off the chip and onto power, land, and time, and those do not scale on a software schedule.

What to watch, and what it means for you

The next two weeks will tell you whether this was a turning point or a busy stretch.

Watch the SpaceX listing on June 12. The first marquee AI-era company to go public sets the emotional tone for the ones lining up behind it, and whether it holds its opening price will say a lot about how much patience the public market really has.

Watch Apple’s developer conference on June 8, and specifically whether Apple lets outside AI models answer questions through Siri. If it does, it cracks open a door for app builders. If it keeps everything in-house, that door stays shut.

Watch for OpenAI’s expected filing, which would put all three AI giants on the public runway together.

And watch the Fed on June 16 and 17, and the next inflation reading before it, because the cost of money is the tide every other plan floats on.

If you take one thing from the week, take this. The companies that grew up as private stories are about to be handed to public scorekeepers who can walk away at any moment, and that same demand, show me the numbers and not the narrative, is rolling downhill to every founder who now has to prove the work instead of describe it. The cheap-money decade rewarded the best storytellers. The one starting now will reward the people whose stories were already true.

Two lines that tracked each other for two decades, US job openings and industrial sales, split apart in November 2022, the month ChatGPT launched. Sales climb to records. Openings fall. The conclusion arrives without anyone having to argue for it. Machines are doing the work, the money keeps flowing, and the people are being left behind.

The video ends at the eight-minute mark with a pitch for a leveraged-long quant trading model “only available this week.” That is what the preceding seven minutes were built to deliver. The charts are persuasion, assembled to sell a subscription, which is reason enough to read them closely before forwarding them to your LPs as evidence of anything.

I read them closely. Some of the claims hold. Several of the charts are doing visual tricks, one of them is stamped with a number its own axes contradict, and one cannot show the thing it is being used to prove. The story underneath the theatrics is the one that matters, and it has already shown up in venture as the sharpest repricing of a business model in twenty years. Here is the honest version, and what it means for the people building, funding, and backing companies right now.

Read the charts before you trust them

Start with the one being shared most. It is also one of the few in the deck that holds up, which makes it the right place to begin.

S&P 500 Index and US Unemployment Rate, inverted. Source: Bloomberg Finance L.P., Bravos Research.

The instinct is to wave this away because the unemployment axis is inverted. Resist it. Inverting unemployment is the standard way to show these two series together, and it is the right call here, because stocks and unemployment normally move in opposite directions. When the economy is healthy, the market rises and joblessness falls, so on an inverted axis the two lines climb together. For two decades they did exactly that, through the 2001 recession, the 2008 collapse, and both recoveries. The reason the chart is worth your time is that the lines have stopped climbing together. The S&P has run to records while unemployment has drifted up, and a relationship that survived two recessions has come apart. That is not a presentation trick. That is the signal.

The harder question is what is causing it, and there are three answers that all point in the same direction. The first is that the S&P 500 has stopped being a gauge of the American economy. The top ten companies now make up roughly forty percent of the index, the highest concentration on record, and the Magnificent Seven alone are about thirty-five percent, up from twelve percent in 2015. Most of those names are the AI infrastructure winners, and their value tracks global compute demand far more than it tracks US payrolls. When the index prints a record, that is now largely a statement about ten labor-light companies, not about the part of the economy that does the hiring. The second answer is margin expansion. The corporate sector is producing record revenue and profit while adding very few net workers, which is the same thing the job-openings chart shows from a different angle. Firms are converting higher output per employee into earnings rather than into headcount. The third answer is the one the video wants you to land on, direct AI substitution, and it is real but smaller than the chart implies. The displacement that shows up cleanly in the data is concentrated among entry-level workers, which I come to below, and it is not yet large enough to move the national unemployment rate by itself.

The three answers matter far more together than apart. Whether the tether between corporate value and labor demand is loosening because the index has become a concentrated bet on labor-light giants, because efficiency is lifting margins without hiring, or because AI is quietly removing the bottom rung, the underlying fact is identical. The link between what the corporate sector is worth and how many people it needs to employ is weaker than it has been in living memory. That is the real content of this chart, and it is structural rather than cyclical, which is what makes it a bigger deal than a single quarter of data.

One caution against over-reading it. Stocks lead the cycle and unemployment lags, so the two diverge around every turning point, and the move in unemployment so far, from a low near 3.4 percent to 4.3 percent, is modest and still historically strong. This is not a recession alarm. It is evidence that the machinery connecting markets to the workforce is changing, which is slower and more consequential than any recession call.

The discipline is needed on the rest of the deck, because the reassurance the video reaches for next does not survive contact with the data. It leans on two examples. One is wrong, and one is built on a chart that contradicts itself.

The Excel example first. The video says Microsoft Excel was released in 1993 and accountants thrived anyway. Excel shipped in 1985 for the Macintosh and reached Windows in 1987. The date is off by close to a decade, and the deeper problem is that the accounting case runs the wrong way for the argument. The Bureau of Labor Statistics projects employment of bookkeeping, accounting, and auditing clerks to fall six percent between 2024 and 2034, and it says plainly that software automated their tasks so the same work gets done with fewer people. Higher-skill accountants grew, modestly. Automation did hollow out the routine clerical layer. That is the displacement story, offered as a refutation of it.

The radiologist example is the one with the self-contradicting chart.

Number of Radiologists and Average Income in the US. Source: Apollo, Bravos Research.

The underlying point is sound. Geoffrey Hinton said in 2016 that we should stop training radiologists because deep learning would surpass them within five years, and instead the field grew and now faces a shortage. The Mayo Clinic’s radiology staff grew fifty-five percent over the period. So far so good. Now look at the chart’s own axes. The number of radiologists runs from about 31,500 to about 37,200, an increase of roughly eighteen percent. Average income runs from about $345,000 to about $565,000, an increase of roughly two-thirds. The chart is then stamped with a callout reading “2x Increase,” which neither line comes close to supporting. Medscape’s compensation surveys put radiologist pay near $400,000 in the mid-2010s and around $500,000 to $526,000 by 2024 and 2025, an increase on the order of a third. The growth is real and the direction is right. The doubling is invented, printed on top of data that says otherwise.

Then the Amazon chart, which is the clearest case of a graphic that cannot show what it is asked to show.

Amazon Worldwide Employment. Source: Amazon Earnings Report, Bravos Research.

The video says Amazon’s hiring “came to a complete standstill” after 2022 while the company spent “trillions” replacing workers with AI. The chart shows worldwide headcount rising explosively to a peak near 1.62 million around 2022, dipping, and recovering to roughly 1.55 million. That is a plateau after a once-in-a-generation hiring spree, not a standstill, and headcount actually rose again in 2024. More to the point, this is total worldwide employment, which is overwhelmingly warehouse labor. Amazon’s AI-driven cuts landed on its corporate workforce, about 350,000 people, where it eliminated roughly 14,000 roles in October 2025 and said advances in AI had raised its confidence in running leaner. Fourteen thousand corporate cuts are invisible inside a 1.5 million headline dominated by fulfillment centers. The chart they chose to prove the corporate-AI story is the one chart that structurally cannot show it.

One correction to my own earlier read, in the interest of accuracy. The “trillions” figure is wrong, since Amazon’s annual capex runs in the low hundreds of billions. But the claim that hyperscaler capex is approaching the entirety of operating cash flow is fair, and their own chart shows it.

Capital Expenditure as a % of Cash Flow From Operations. Source: Bloomberg Finance L.P., Bravos Research.

By early 2026 Amazon and Meta are spending roughly ninety percent or more of operating cash flow on capex, with Google not far behind and Microsoft climbing fast. That is the real and important fact buried in a video full of unreal ones.

The story underneath the theatrics

Strip out the tricks and a genuine structural shift remains, and it is about capital, not headlines.

Real Non-Residential Fixed Investment. Source: BCA Research, Bravos Research.

High-Tech Versus Low-Tech Capital Spending in Nominal GDP. Source: Yardeni Research, National Bureau of Economic Research, Bravos Research.

Investment in information processing has gone vertical while investment in everything else has flatlined, and high-tech capital spending is surging as low-tech spending sits near zero. This is the part outside analysts confirm without a sales motive. In the first half of 2025, AI-related capital expenditure contributed more to US GDP growth than all consumer spending combined, the first time data center buildout outweighed the American shopper as the engine of the economy. The five largest cloud and AI infrastructure companies have committed to somewhere between 660 and 750 billion dollars of capex in 2026, close to double the 2025 figure, with Amazon alone planning roughly 200 billion. Capital is being pulled into a narrow set of assets, chips, data centers, and the power to run them, and away from housing, transportation, and the broad industrial economy. Data centers employ very few people once built. That is the actual mechanism behind the chart everyone is sharing.

The labor effect is real too, and it is more specific and more uncomfortable than a macro divergence can show. The strongest evidence sits in payroll microdata, not in the video. Stanford economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, working with ADP records covering tens of millions of workers, found that early-career workers aged 22 to 25 in the most AI-exposed occupations have seen employment fall by roughly thirteen percent since late 2022, even after controlling for firm-level shocks. Older workers in the same roles held steady or grew. The damage is concentrated at the entry level, where the work is codified and learnable from text, and it is invisible in headline unemployment because it shows up as jobs that are never created rather than workers who are fired.

The honest counterweight, which the video skips and which I will not, is that the aggregate labor market is fine.

Unemployment Rate and Initial Jobless Claims. Source: Bloomberg Finance L.P., Bravos Research.

Initial jobless claims are near multi-decade lows, which is the chart admitting that companies are not actually firing people in large numbers. The May 2026 jobs report, released after the video, made the point harder. Payrolls came in at 172,000 against a consensus near 80,000, and unemployment held at 4.3 percent for a third straight month. The aggregate is resilient. The stress is in the composition, at the entry level and among the long-term unemployed, not in the level. Hold those two facts together. Capital is concentrating into a handful of AI bets, and the labor effect so far is concentrated at the bottom of the experience curve.

Where this already hit venture

For a fund whose thesis includes B2B SaaS, the abstract story stopped being abstract in February 2026.

In the first week of that month, software stocks shed more than a trillion dollars in market capitalization, by Forrester’s count, in a stretch the firm and others started calling the SaaSpocalypse. The trigger was the arrival of genuinely capable autonomous agents, including Anthropic’s Claude Cowork and Claude Code and OpenAI’s agent mode, and the market’s sudden conviction that per-seat software pricing is exposed when one agent does the work of several licensed users. The iShares Expanded Tech-Software ETF fell more than twenty percent in the first quarter. Salesforce and Workday dropped roughly thirty percent, Adobe traded near half its early-2025 level. The selloff recurred in April when Anthropic shipped managed agent hosting and the infrastructure names that sold those services separately, Cloudflare, Akamai, DigitalOcean, fell double digits in a day.

This is the same force the Bravos charts are pointing at, expressed in the one place a venture investor cannot look away from. Enterprises are moving budget toward AI infrastructure and agentic workflows and away from seat-based subscriptions. The market has decided that recurring revenue tied to human headcount is no longer the safe, compounding asset it was treated as for two decades. Whether the repricing is an overshoot or a permanent revaluation is the live debate, and reasonable people are on both sides. What is no longer debatable is that “we sell software by the seat” is a sentence that now carries risk it did not carry eighteen months ago.

Everything below follows from that.

For founders

The cost of building collapsed, and that cuts both ways. The same forces letting Amazon and Walmart run leaner let a seed-stage company reach real revenue with a fraction of the team the same milestone required five years ago. Small teams reaching seven and eight figures of revenue without raising are no longer anomalies. The practical effect is that the round you need to hit a given milestone is smaller, the team you hire against it is more senior, and the path to default-alive is shorter. The founder raising a large seed to hire twenty people is competing with a founder who raised a third of that and runs the same playbook with agents and five operators.

The junior-leverage model is over. The old way to scale was to hire cheap junior talent and supervise it. The Stanford data is the warning that the market is repricing exactly that labor, and your own company will feel it before your customers do. Staff with fewer, more experienced people and more tooling, and assume your competitors will too.

Pricing has to move with the buyer. Enterprise customers are revolting against paying for hundreds of licenses when a handful of agents do the work, which is the mechanism behind the SaaSpocalypse. Monetization is shifting toward consumption and outcomes, charging per resolved ticket or completed audit rather than per logged-in human. If your revenue model assumes seat growth that tracks your customer’s headcount, you are modeling a number your customer is actively trying to shrink.

Defensibility moved, and this is the screen that matters most. If a model release can rebuild your product, then having built it is not a moat. The durable positions are distribution, proprietary data, deep workflow integration, switching costs, and regulatory standing. A pitch that amounts to “we built this with AI” describes a feature with a short half-life, because the thing AI made cheap for you to build it made cheap for the next four teams as well. The question to answer in your own deck is what survives the next capability jump.

For VCs

The capital picture is a barbell and the middle is thin. Enormous sums flow into AI labs and infrastructure at the top while the long tail of early and growth stage competes for what is left. For a pre-seed and seed manager this is threat and opportunity in one motion. The threat is that the Series A bar rises, because the comparison is no longer the other startup, it is whether an incumbent could do this with agents, and graduation rates fall. The opportunity is that entry prices for unglamorous categories outside the AI narrative stay reasonable while everyone crowds the hyped end.

Underwriting has to change with the margin structure. Classic SaaS earned eighty percent gross margins because code was written once and copied for free. AI-native products carry real compute and token costs in cost of goods sold. Top-line ARR is no longer a clean proxy for quality. Net revenue retention and contribution margin after compute are the numbers that separate a durable business from a venture-subsidized demo, and the SaaSpocalypse is the market enforcing that distinction in public.

The selection question sharpened into one test. Are you backing companies that AI makes possible, or companies that AI makes obsolete. A product the next model delivers for free is a capability-timing bet that the timing runs out on, and it looks identical to a great company right up until it does not. That is the failure mode worth screening hardest for.

Markups are not returns, and this is the cycle where the distinction bites. AI-adjacent positions are being marked up on the same enthusiasm carrying the public indices, and the historical rhyme is the telecom and dot-com overbuild, where the infrastructure thesis was correct and most of the early equity was still wiped out. Taking partial liquidity through secondaries on inflated positions is discipline, not a lack of conviction.

The one lesson to refuse is the video’s own. It exists to sell market timing, and venture is the asset class where timing the macro is least available to you. You cannot call whether the capex cycle runs a decade like the railroads or cracks in eighteen months like the fiber buildout. Deploy through it with disciplined entry prices and let the volume of bets carry the uncertainty, because the honest read on which way this breaks is that nobody knows.

For LPs

The liquidity problem is no longer a forecast, it is a measured fact, and it is severe. Net cash flows from venture funds to LPs have been negative by about 169 billion dollars since 2022, per PitchBook and NVCA data. The median 2017-vintage fund, now eight years old, has returned roughly a quarter of paid-in capital, with median DPI around 0.28x. Newer vintages are worse, with only single-digit percentages of 2021 funds having returned any cash at all after three years. Distribution rates have run at single-digit percentages of net asset value for eight straight quarters against a decade average near seventeen percent. This is the backdrop against which every fund you are offered is being raised.

The diversification you think you have is thinner than it looks. If AI capex is simultaneously driving GDP, dominating the public indices through a handful of names, financing itself through private credit on data centers, and setting valuations across venture, then a portfolio holding all four is not diversified across risks. It is four expressions of one bet on the AI capex cycle continuing. The venture allocation that was sold as uncorrelated is now correlated to the same names that move the S&P, and the SaaSpocalypse showed how fast a single product launch can reprice the lot.

Secondaries have gone from a workaround to plumbing, and you should use them as such without expecting them to fix the underlying problem. The secondary market has grown past 150 billion dollars in annual volume, Goldman Sachs paid up to 965 million for Industry Ventures to build a position in it, and continuation vehicles are now a standard tool for GPs who want to hold AI assets past fund life. PitchBook’s own read is blunt, that secondaries provide a liquidity lifeline but lack the capacity to resolve the DPI crisis. They rebalance the problem, they do not retire it.

Weight realized cash over paper, and entry discipline over theme exposure. Every manager you meet is now an AI manager, so the question is not whether a fund is exposed to AI, it is the price at which it entered and whether it holds reserves to defend positions if the cycle turns. A fund deployed at the top of the valuation cycle into AI-adjacent companies at inflated prices carries vintage risk that no amount of thesis eloquence removes (this is less of an issue in Pre-Seed where we mostly invest). With public exit windows thin, the manager who can show a path to actual distributions is worth more than the one with the prettier markup.

Finally, hold both skeptical positions at once. The base rate on technology-displacement panic is poor. The radiologists are still employed, the people who said otherwise are on record, and a video that prints “2x” over a chart showing two-thirds should lower your trust, not raise it. At the same time, the entry-level data is real, the SaaSpocalypse was real cash leaving real portfolios, and complacency is its own mistake. The position that survives both is not a forecast. It is high uncertainty, which is the case for spreading bets across volume rather than concentrating them on whichever narrative is loudest this quarter.

What to take from the charts

The divergence is worth your attention, and not for the reason it is being shared. The labor story it implies is weaker than it looks, propped up by an inverted axis, a fabricated multiple, and a chart that cannot show its own claim. The capital story underneath is stronger than the people selling it seem to realize, and it is the one with teeth for anyone managing a fund. Money is concentrating into a narrow set of AI bets, the cost of building has fallen far enough to reshape what a startup needs to raise and who it hires, the labor effect that is real is hitting the on-ramp where it is hardest to see, and the whole thing has already reached venture through the repricing of seat-based software.

None of that resolves into a trade you can put on this week. It resolves into discipline. Price your entries, weight your bets across the uncertainty, take some chips off the table when it is offered, and treat any chart that ends in a sales pitch as the beginning of your own work rather than the end of it.

First, the phrase. Recursive self-improvement is the point at which an AI system can design and build a better version of itself with no human in the loop. The model writes the next model’s code, runs the experiments, reads the results, and decides what to try next, on its own. Anthropic is careful to say we are not there, and that it is not guaranteed to arrive. What they are claiming is narrower and harder to wave away. They are saying the loop is closing one stage at a time, and they have internal numbers showing how far it has already closed.

Which numbers to believe

The headline most people will repeat is that Anthropic engineers now ship roughly eight times the code per person they did a couple of years ago. Ignore it. Anthropic itself flags that lines of code measures volume, not value, and that the real productivity gain is almost certainly smaller. A model that writes verbose code looks more productive on this chart while making the codebase worse. So does a model cleaning up four years of deferred junk in an afternoon. Treat the 8x as a mood, not a measurement. High confidence it overstates the truth.

The numbers worth your attention are the ones that are hard to game. The first is task length. A research group called METR measures the longest task a model can finish on its own with reasonable reliability, and that length has been roughly doubling every four months, faster than it used to. In March 2024 the best Claude could handle a software task that takes a person about four minutes. A year later, ninety-minute tasks. A year after that, twelve-hour tasks, with the unreleased internal model running past sixteen hours and bumping against the ceiling of what METR can even measure. Lines of code can lie. The clock cannot. If that curve holds, work that takes a skilled person a full week comes into range during 2027. That is the trend to watch, and I would put it at high confidence in the near term and moderate confidence past a year, because every exponential eventually meets a wall and we cannot see this one yet.

The second is the success rate on genuinely open-ended problems, meaning problems with no clear spec where even the engineer does not know what the answer looks like. Anthropic reports that rate hit 76 percent in May 2026, up fifty points in six months. The example they give is a model handed a live incident, tens of thousands of training jobs crashing, given little more than cluster access, that isolated one obscure debugging flag and shipped a fix in two hours where a person would have spent days. That is not autocomplete. That is debugging under uncertainty, which is most of what senior engineering actually is.

The rest of the evidence rhymes. A standard software-engineering benchmark went from near zero to solved in two years. A benchmark for reproducing published research went from one in five to solved in fifteen months. More than four out of five lines merged into Anthropic’s own codebase are now written by Claude, up from low single digits before their coding agent launched in early 2025. Read those as confirmation, not as separate miracles. They all point the same way.

The job that is left

Here is the part that should hold the attention of anyone whose living comes from judgment rather than output, which includes my entire industry.

Anthropic is candid that the human role has narrowed to a single function at each stage. Humans no longer write the code, they review it. Humans no longer run the experiment, they choose which experiment is worth running. The doing has collapsed toward zero cost in human time. What remains is taste. Knowing which problem matters, which result to trust, when an approach is a dead end. They call it research taste and they name it as the last real moat between a capable assistant and a system that could replace its makers.

And then, in the same piece, they put that moat on the clock. They note that taste might be just another capability that models fail at for a while and then suddenly get good at, the way they eventually learned to explain why a joke lands or pass a theory-of-mind test. They have an early measurement to back the worry. On a set of real research sessions, they asked models to pick the next step, and the best model in November 2025 beat the human choice 51 percent of the time, rising to 64 percent by April 2026. You should discount this one, and the article tells you how. They deliberately chose moments where the human’s move had room to improve, so it is not a fair fight. On a control set where the human’s move was already strong, the models won only about a fifth of the time. So taste is not falling yet. But the people with the most information in the world just told you it is the next thing they expect to fall.

Sit with what that means if you price judgment for a living. The pitch of every active venture investor, every stock picker, every senior analyst, is some version of “I see what others miss.” That is a taste claim. If taste compresses the way coding did, the premium you charge for it compresses with it. I am not predicting that this year. Moderate confidence it starts to bite within three to five years for narrow, well-defined judgment, and low confidence on the open-ended, cross-domain judgment that separates a good seed investor from a lucky one. But the direction is not in dispute, and the people running the experiment are not the ones with an incentive to hype this particular finding.

Why I am not panicking, and why that is not reassuring

I have spent five years building a firm around the bet that the doing is cheap and the picking is overrated. The strategy is breadth. Cover selectively, but broadly, rather than agonize over three names, because at the earliest stage the variance in outcomes swamps anyone’s ability to forecast them. This article reads like a long external validation of that view. If implementation is nearly free and even expert judgment is unreliable enough to be beaten on a coin flip in cherry-picked cases, then conviction is worth less than coverage and speed.

That is the comfortable reading, so I should attack it. The uncomfortable version is that cheap doing and compressing taste do not protect a volume strategy, they commoditize it. If access and speed are my edge, and access and speed are exactly what an army of agents grants every other allocator too, then breadth stops being a moat the moment everyone can run it. The thing that survives is not a strategy at all. It is the stuff that did not get cheaper. Relationships a founder chooses to take money from. A brand that earns the first call. The willingness to act when the model says wait. None of that lives on the benchmark.

The law nobody quotes

The single most useful idea in the piece is buried near the end, and it is borrowed from computing. Amdahl’s law says that speeding up one part of a process only helps until the parts you did not speed up become the limit. Anthropic has already hit it inside their own walls. They taught the machines to write code faster than humans can review it, so review became the new bottleneck. The constraint did not vanish. It moved.

That migration is the whole investment thesis hiding in this report, and they nearly walk past it. When engineering goes to roughly free, value pools wherever the bottleneck lands next, and the bottleneck lands on everything that refused to get faster. Their own cyber program is the clean example. The internal Mythos model found more than ten thousand serious software vulnerabilities in its first weeks, and the binding constraint in defense flipped overnight from finding holes to patching them fast enough. Finding got automated. Patching, which touches deployment and humans and downtime, did not. If you want to know where the next decade of company-building money goes, stop looking at what AI makes cheap and start looking at what stays expensive next to it. Distribution. Trust. Regulatory throughput. Anything physical. Anything that requires a human to say yes.

The part where they ask for help they know they will not get

The closing section proposes that frontier labs keep the option to slow down or pause, and that Anthropic would do so if rivals verifiably did the same. Read the conditions and you have read the obituary of the idea. A training run is easier to hide than a missile silo, the inputs are general-purpose, and whoever keeps going while others stop inherits the lead. They cite the decades it took to stand up nuclear arms verification and then admit, in the next breath, that we do not have decades. The honest translation of the proposal is that a verifiable pause is close to impossible, a unilateral one only changes who finishes first, and so the default is that nobody pauses. I do not read that as cynicism on their part. I read it as the most useful sentence in the essay, said quietly so it does not spoil the mood.

So we are left with the ratio I started with. The afternoon I spend deciding whether the memo is right is the job now, and the report I just read is a careful, well-sourced argument that even that afternoon is on a timer. I do not think it runs out this year. I think anyone betting their career on judgment should assume the premium they charge for it is going to thin, and should be building the relationships and the brand and the nerve that no benchmark can saturate while the charge for cleverness still holds.

The doing got cheap. Be very clear-eyed about what you are selling once it does.

Build something interesting. Pitch investors. Get funded. Grow fast.

But anyone who has actually raised money knows the process is rarely that clean. Great companies get passed on. Mediocre companies get funded. Investors give conflicting feedback. One VC says the market is too small. Another says it is too crowded. One says come back with revenue. Another says the revenue is the wrong kind.

For founders, the whole thing can feel arbitrary.

Charlie O’Donnell wants founders to understand something uncomfortable but useful: venture capital is not a fairness machine. It is a financial product with a very specific job.

Charlie has seen that machine from almost every angle. He started on the institutional LP side at the General Motors Pension Fund, evaluating venture funds after the dot-com crash. He later became the first analyst at Union Square Ventures, helped First Round Capital open its New York office, and eventually launched Brooklyn Bridge Ventures, the first VC fund based in Brooklyn. Across his career, he wrote first checks into more than 100 companies and built a reputation as one of New York’s most accessible early-stage investors.

Now, after stepping away from active fund investing, Charlie is helping founders understand what VCs often do not say directly. His book, Founder Unfriendly: What Investors Won’t Tell You About Getting Funded, is written for the founders who are not already famous, not already backed by elite networks, and not already surrounded by five venture-backed friends who can review every pitch deck, co-founder decision, and investor intro. In other words, most founders.

Subscribe now

Venture Capital Is Not the Same as Business Validation

One of the biggest mistakes founders make is assuming that a VC pass means the business is bad.

That is wrong.

A business can be excellent and still be a terrible fit for venture capital. Charlie points out that a company generating $5 million in free cash flow every year might be a phenomenal business for the founder. But if it cannot become large enough to return a major venture fund, it may not fit the VC model.

This is where founders often misunderstand the investor’s job.

VCs are not simply asking, “Is this a good business?”

They are asking, “Can this become big enough to return our fund?”

That distinction matters. A profitable, durable, founder-owned company may be far more attractive than a venture-backed company forced onto an unnatural growth path. But if a founder walks into a VC pitch without understanding the risk-return model of venture, they can mistake rejection for judgment.

It is not always judgment. Sometimes it is just fund math.

The Best Founders Do Not Pitch Small

Charlie’s strongest fundraising advice is also one of the most counterintuitive: founders often lose investor interest because they pitch what they can confidently promise, not what the company could become.

That instinct is understandable. Many founders do not want to overstate. They do not want to look naïve. They do not want to project $200 million in revenue and then fall short. This is especially true for founders who feel they are under more scrutiny because of their background, gender, race, or lack of insider status.

But according to Charlie, that caution can backfire.

VCs are not looking for a conservative promise. They are looking for fund-returning potential.

That does not mean founders should lie or inflate numbers. It means they need to clearly explain what happens if the company works. What does more capital unlock? What would it mean to double the sales team, expand into more cities, launch faster, or capture the market before competitors do?

Charlie frames it simply: fundraising is not a promise conversation. It is a potential conversation.

Founders who only pitch the safe version of the company often make the opportunity sound too small. The investor may never see the upside case because the founder never actually says it.

Control the Meeting or the Meeting Controls You

Another mistake founders make is letting investors take over the pitch.

VCs will ask questions. Some will be useful. Some will be distracting. Some will pull the conversation into downside risk before the founder has even explained the upside. If the founder simply follows every question wherever it goes, the meeting can become fragmented and defensive.

Charlie argues that founders need to bring structure.

That does not mean being obnoxious or overbearing. It means setting the frame.

A founder might start by saying: here are the three things that get people excited about this company. If I convince you of these three things, would this be worth spending more time on?

That structure changes the meeting. It gives the founder a clear agenda. It lets the investor opt into the logic. And at the end, the founder can bring the conversation back to the original case: did I convince you of the things that matter?

This is not just presentation polish. It is a signal.

Investors are not only evaluating the business. They are asking whether this founder can recruit great executives, close impossible customers, handle skeptical partners, and keep control of high-stakes rooms.

A founder who cannot control a VC meeting may struggle to convince the investor they can control much harder rooms later.

Networks Are an Unfair Advantage

Charlie is blunt about one of the least fair parts of fundraising: networks matter enormously.

A founder with five venture-backed friends has a major advantage. Those friends can review the deck, make warm intros, explain how firms think, help evaluate a VP of Sales candidate, and translate confusing investor feedback.

That knowledge is not evenly distributed.

Founders outside those circles often have to work much harder to get the same information. They may be just as capable, but they are operating without the same insider map.

This is one of the reasons fundraising can look meritocratic while quietly favoring people who already know the rules.

Charlie’s advice is not to complain about the unfairness. It is to recognize it and deliberately build the network anyway. The best founders are often better not because they were born with perfect judgment, but because they have better access to people who sharpen that judgment.

That is not fair. But ignoring it is worse.

The AI Era Has Raised the Bar

One of Charlie’s clearest reversals is around product.

Years ago, he was willing to back founders before they had a product. In some cases, that made sense. If the team was exceptional and the idea was complex, getting in before the product existed could create a pricing advantage.

But that logic has changed.

In the AI era, the bar for building something has dropped. For many software startups, showing up with no product is no longer a sign of being early. It can be a sign of adverse selection.

If the tools to build are faster, cheaper, and more accessible, then investors expect more. A founder pitching a software company without even a basic product now faces a harder question: why not?

For founders, the implication is direct. The old “idea-stage” pitch is weaker than it used to be. Unless the company is in deep tech, hard science, or another category with real technical barriers, investors increasingly expect proof that the founder can turn insight into something tangible.

Fundraising Is Winnable, But It Is Not Fair

The central lesson from Charlie’s career is not that VCs are villains or founders are naïve. It is that both sides are operating inside a system with incentives that are often misunderstood.

VCs need outlier outcomes. Founders need to understand the type of outcome they are pitching. Investors are pattern matching. Founders need to know which patterns help them and which ones hurt them. The process is biased toward networks, confidence, and clarity. Founders who lack those advantages need to build them intentionally.

The danger is pretending the process is fairer, more rational, or more transparent than it really is.

Charlie’s message to founders is not comforting. It is more useful than that.

A VC pass does not mean your company is bad. A fundraise is not a referendum on your worth. A good business is not always a venture-backable business. And if you are pitching investors, your job is not to present the smallest thing you can safely defend.

Your job is to make the upside impossible to miss.

Because venture capital does not fund what is merely sensible.

It funds what could become enormous.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:
03:55 — Surviving the Dot-Com Crash and Negative Returns
06:29 — What LPs Don’t See About Venture Capital
09:39 — Why VCs Are Still Middlemen in the Startup Ecosystem
11:33 — Lessons from Being the First Analyst at Union Square Ventures
14:02 — Building a Network Without Money or an Ivy League Background
17:10 — Creating Access Through Community and Events
19:57 — Joining First Round Capital After a Failed Startup
20:31 — Pitching During the 2008 Financial Crisis
21:01 — Helping Spark the Foursquare Funding Race
22:28 — Why New York Needed a Different VC Playbook
24:26 — GroupMe, SinglePlatform, and Early Wins at First Round
25:33 — Price Sensitivity vs. Price Takers in Early-Stage VC
28:05 — Why One Lucky Deal Is Not an Investment Strategy
32:15 — Leaving First Round to Launch Brooklyn Bridge Ventures
34:21 — Why Charlie Walked Away From Active Fund Investing
37:09 — Writing Founder Unfriendly for the 99% of Founders
39:20 — Why Good Businesses Still Get Rejected by VCs
41:00 — Pitching Potential Instead of Conservative Promises
45:35 — Why Fundraising Is a Potential Conversation
46:10 — What Founders Can Learn From Parenting a Small Child
47:30 — Why Every Slide Needs to Scream Fund-Returning Outcome
48:30 — Team, Market, and Traction as the Core Pitch Narrative
50:48 — How Founders Can Redirect Bad Investor Questions
53:28 — Controlling the VC Meeting Without Being Obnoxious
55:47 — Why Founders Should Read Founder Unfriendly
56:41 — The One Deal Charlie Wishes Hadn’t Fallen Through

Transcript

Brian Bell (00:01:23): Hey everyone, welcome back to the Ignite Podcast. Today we have Charlie O'Donnell on the mic. He spent 20 years inside rooms where funding decisions actually get made. First analyst at Union Square Ventures, then helping First Round open its New York office, where he sourced GroupMe, Single Platform Moat, Backupify, In 2012, he founded Brooklyn Bridge Ventures, the first VC fund based in Brooklyn, and wrote first checks in over 100 companies, including Hungry Root, Petal, Bridget, Gotenna, and The Wing, and others, building a reputation as the most I'm looking forward to that conversation and running the next NYC community. This spring, he just published Founder Unfriendly, What Investors Won't Tell You About Getting Funded, which hit number one in the new release in the VC category and carries a cover blurb from Brad Feld. Very cool. Thanks for coming on, Charlie.

Charlie O’Donnell (00:02:22): Thanks for having me. I appreciate it.

Brian Bell (00:02:23): Yeah, so I'd love to get your origin story. It's so fun to kind of hear VC stories because they're so different. But like, you know, what is your background?

Charlie O’Donnell (00:02:31): So actually, my first job in venture capital was on the institutional limited partner side. So if you're out there pitching VCs, and it's frustrating, keep in mind that they also have to pitch to raise their funds. They don't necessarily I have to do it as often as you do, but we were the big pool of money behind a lot of the top tier names that you've heard of. GM had been investing in venture as an asset class since the 70s. So Union Square Ventures, when they raised their first fund, came to pitch to me back in 2004. So that was my entree into venture is funding funds and then eventually joining a fund.

Brian Bell (00:03:13): That's amazing. How did that, I mean, was that just right off the college campus? Like how do you land a job like that? That's pretty amazing.

Charlie O’Donnell (00:03:20): It was kind of funny. It was actually a high school internship. So my high school, which is pretty smart, actually, after we had applied to our colleges, they basically kicked us out because, you know, you've you've applied to your schools. you're just kind of hanging around campus like not really paying attention sort of the senioritis thing and so they're like okay we got to get you guys out of here because all you have to do is pass and so you're just kind of mailing it in so you had a choice of either doing a community service project I'm what luck that's super glamorous keep in mind the timing in 2001 I saw 10 negative quarters of performance before I ever saw a positive one I mean this was just the period of shutting down dot-com and depending on when you had raised your last round, you either had like two years of money left or two months of money left, but you definitely weren't a viable business. So the VCs would sort of come in and every quarter they'd be like, okay, we did a triage and we think like A third of this portfolio is going to make it and return your money and a third is kind of on the fence and a third is just, you know, garbage. And then the next time we met them, we're like, okay, well, you know, actually it's a third, a third, a third. And after three or four rounds of that, they're like, I don't think our 99 vintage fund is going to return capital. Let's look towards the future. And so it was a really tough space at the time.

Brian Bell (00:05:05): what a tough conversation right yeah yeah for sure and I think there's some really famous examples of I think like it was like benchmark around that time just decided like we're not going to take management fees for two years and we're going to deploy that and save the fund basically there's a lot of famous examples of that but what a tough conversation with your with your LP is hey you know that money you gave me I'm not going to be able to return it well the thing that helps with some

Charlie O’Donnell (00:05:25): of them though is you think about a fund like Excel I think there's like their 1995 fund or 1994 was like a 25x returning fund. So if you return 25x and then even like a really terrible fund.

Brian Bell (00:05:39): You have a 1x fund or whatever. Right. Like even if it's less than 1x. You're still averaging 12x.

Charlie O’Donnell (00:05:44): You're still way ahead, right? Yeah. So you can take a mulligan on one fund, maybe not two.

Brian Bell (00:05:51): Yeah, you got caught by the market there.

Charlie O’Donnell (00:05:53): Right. But also the timing makes it really interesting. So what wound up happening is normally in any given year, maybe like a quarter or a third of your portfolio would come back to market. but everyone stopped investing at exactly the same time which meant everyone came to market again roughly around the same time so in 2004 something like plus or minus you know 18 months around 2004 like 85% of our portfolio came back to market so we had the unique opportunity to either say we want out of venture like literally we're just not doing any new deals and I think Tia Kreff actually completely pulled out of the asset class We want to double down, right? Like we could actually pull into these funds as other people are pulling out or we're still into venture but we don't like these managers and we're going to switch out our managers and so there are a lot of emerging funds like Union Square Ventures, Spark Capital, Emergence. They were all sort of born around this time period of, you know, and several of those firms had people who came from other firms. Todd Degris came from Spark and there was a whole reshuffling of the asset clients. that resulted in a lot of new funds around that time. So it was actually really quite interesting to be an analyst at a institutional fund looking at the whole space and trying to figure out like, is this still viable? What names do we want to go into?

Brian Bell (00:07:18): So working at a pension fund, what did watching this flow of institutional LP capital inside teach you most of that CGPs never see?

Charlie O’Donnell (00:07:28): So I have to admit, I'm a little skeptical when people say, well, the returns haven't been great. Will LPs pull out? Because the one thing when you're looking across asset classes is if you think somebody is potentially a little skeptical about venture, you have to ask yourself, where else are they going to put the money? Right? Because if you're a manager... Right, exactly, right? And so, you know, you look at valuations and public equities, right? You're not going there, right?

Brian Bell (00:08:01): Late stage, even late stage private markets now, the valuations are pretty crazy. For sure, right? Like they were in 21.

Charlie O’Donnell (00:08:08): You look at interest rates and real estate, like there's just not a lot of places where you could actually tell yourself a story that if you have good managers to back, that you can make this outsized return. And so, you know, this is where the venture train keeps on running and why some of the bigger funds continue to return capital to continue to raise capital. is because they have big funds, they have prior outperformance. And so if you have an allocation into the union squares and lightspeeds and, you know, and reasons and benchmarks and wherever, you're probably not going to give that up, actually, because you don't have another opportunity that you can tell yourself that you're going to get like these kinds of returns out of. So that's one. I think the second thing And when I moved from the General Motors Pension Fund to Union Square Ventures and got to talk to founders about who they wanted to raise their A and what they thought of different funds, I realized maybe today it's a little bit different because there's more transparency in the market through social media and newsletters and LinkedIn and all this other stuff. We were so disconnected from that Next great entrepreneur who's choosing where to go pitch their fund and you know our perceptions of the market were years like like almost a generation behind actually because LPs don't have relationships with with founders and not just the successful founders, right? Like that next founder, the VP of engineering from Plaid who discovers some infrastructure thing and trying to implement AI across like fintech backends or whatever, like That's the decision maker you need to tap into The Massachusetts State Pension Fund

Brian Bell (00:09:57): They're not talking to that person They also just don't have the infrastructure To screen thousands and not over 10,000 deals Like GPs do I'm actually going through and revamping some of my slides And kind of telling that story of the funnel The deal slow funnel Because we're a very high volume investor Investing in a hundred a year across funds two and three. And, you know, it's very much an education process of like, how are you seeing, you know, 10, 15,000 deals a year, right? How are you like physically able to do that, right? And telling that story and that infrastructure in place. And it's an interesting, like you could almost like say like VCs are kind of like middlemen a little bit. A little bit. Yeah, yeah. It's like you kind of sit in the middle, like maybe if the LPs had all the same infrastructure, if they just took all my systems and like hired an analyst, maybe they can do the investing themselves. But I think there's like also this like persistence of returns in VC because the deal flow and your network gets better over time, right?

Charlie O’Donnell (00:10:55): Well yeah right it's it's sort of like when you get a chance to see what greatness looks like you're part of a rocket ship outcome then you surround yourself with people who also have that experience you may not have been the founder What did you learn at being the very first analyst at Union Square?

Brian Bell (00:11:13): I mean, you're coming in pretty green from the LP side. Like, what did you What did you have to unlearn and what were some surprising lessons in that role?

Charlie O’Donnell (00:11:44): I think the biggest thing is how much of the knowledge was sort of locked up in people. So I came from a world where at the Bloomberg terminal on my desk, I could call up any, you know, trade desk and get the, you know, sell side analyst report.

Brian Bell (00:12:01): Give me your bid ask on this and yeah.

Charlie O’Donnell (00:12:03): Yeah, and the sell side analyst report on, you know, whatever security sector and all this sort of stuff. When I went to USV, I was like, these are sectors that barely even exist yet. And they're being created as we speak. And like, who knows I am a And so that's one of the reasons why I became more public on social, started running events. I had to create the community in New York in 2004 and 2005 that was necessary for me to learn, actually. actually and so I've always been sort of a a community creator and and people connector because it was by necessity right like I need to surround myself with the people I need to learn from so that I can be good at my job it's not the kind of job you can do from behind a desk I mean now post-pandemic AI screening harmonic

Brian Bell (00:13:18): sit in my pajamas all day and take zoom calls basically but yeah right right but at

Charlie O’Donnell (00:13:22): the time you couldn't do it though

Brian Bell (00:13:23): yeah yeah yeah not 20 years ago that's around the same time I lived in New York as well had dreams of you know working on Wall Street and all that stuff and yeah I wish I would have gotten to VC then you know I have my notes you said you carried the privileges of being a white guy but didn't grow up with money or an ivy network you know how did you you talked a little bit about creating that network what do you what do you think it takes today to to kind of build that network from scratch

Charlie O’Donnell (00:13:45): so first of all it takes it takes confidence it you have to first First convince yourself that anybody wants to meet with you. So I think confidence is a big thing. I think you have to act as if and be willing to reach out to people and assume that you can create value for them if you don't have the background.

Brian Bell (00:14:02): Confidence will take you all the way to the presidency is what I like to say.

Charlie O’Donnell (00:14:07): Apparently. Whether it's founded or you've all founded or not.

Brian Bell (00:14:11): Yeah, misplaced confidence. I know the best people. I'm the best. I'm the best at everything.

Charlie O’Donnell (00:14:15): Exactly. You know. I think one confidence hack is that when you create the opportunity for conversation from people, you're actually doing somebody a favor. And so it's not, why does this person want to talk to me? But am I going to create a space for this person to share their ideas? Because, you know, everybody's got to feed the beast every day now, right? Like the prior professional Rolodex that we had in LinkedIn is now, you know, a social network where you need to like look smart every day. And sometimes you wake up and you've got nothing, right? But if somebody comes along and says, hey, I'd like to interview you and I'm going to share these clips and You know we're going to spotlight what you're thinking you're going to have an opportunity to put something out there or an in-person event right like you might get asked a question at an in-person event that like really inspires your thinking changes direction who knows who you might so creating those opportunities for someone is actually a really valuable offer it's not an ask and so I think that is really incredibly important but you know not everybody feels that way right not everybody lives in a kind of abundance world where something good could happen when I take time for you right that's a very like white collar kind of you know up into the right sort of approach right Versus like, hey, if I waste your time, you could have been out on a plumbing job fixing something. And there's like the only upside when you're a plumber is to get paid for your time. And time is money and that's it. And everything's sort of a very limited kind of upside world. So I think that is, you know, the number one approach and the willingness to just ask a lot of people. Like I do this pre-Series A offsite, this one day conference where I get But 14 or 15 check writing series A VCs, partners, select principals, and we literally kidnap them. They're supposed to be there the whole day. They can't walk off after their talk the way typical panelists might might be and so it's a it's a big ask but when we make that ask and we get those people when we get partners from NEA and GV and USB and Lightspeed and all these other folks then founders really want to come to it actually because these folks are going to be here all day right so to get 15 yeah we need to ask about 70 or 80 VCs and does the timing work and are they going to be in New York and can they spare the day and are they convinced this is going to be valuable for them and so like I don't mind getting 50 no's for this event because I know that I'm going to get people and it's going to work out. And we've now done that event 10 times. I'm sort of used to it. But, you know, when you're reaching out for two or three weeks and you're just like, oh, my God, I just I've won for a minute.

Brian Bell (00:17:13): get this thing put together yeah I know yeah it's funny you mentioned you mentioned something there which I want to tug on which is the you know sometimes you wake up and you got nothing you know and there's days as a VC where I wake up I'm like man I really don't want to go to work today I just want to you know cancel everything and just go back to bed but then I meet like some amazing founder you know just working on the coolest thing and I'm like this is my job this is what I literally get to do and and I you know it's kind of like going to a party sometimes you don't want to go to a party And you're just like, I'd rather just stay home, watch Netflix, you know, just want to like just veg out. And then you always go and you kind of meet somebody cool and you're like, I'm glad I went. You always come home from a party like glad you went. And I think like being a VC is like going to a party every day. You're like, sometimes you don't want to go to the party. You just want to introvert. I just don't want to like talk to anybody. My wife's always making fun of me because she's like, why don't you like know like our kids' friends, you know, parents' names. I'm like, this is like my job is to meet people all day long, you know, so I just want to introvert. But I'm always glad I went, you know, to work.

Charlie O’Donnell (00:18:14): Yeah, no, I agree. I agree.

Brian Bell (00:18:16): So tell us the story of First Round. So you opened up the first office there.

Charlie O’Donnell (00:18:20): So yeah, I got hired by First Round. It's pretty random, actually, because there wasn't really a job opening. I had started a company after I left Union Square. and I wasn't particularly good at being an entrepreneur and I kind of muddled through it and we were down to our last negative $30,000 in capital. I maxed out on my credit cards when I landed at first round.

Brian Bell (00:18:45): What was happening financially?

Charlie O’Donnell (00:18:47): so yeah I had started a company and it didn't really I started a company in 2007 worked on it for a couple years didn't really work out tried to pitch in October of 2008 I'm like literally I was in SF yeah I'm literally in SF and there's a VC sitting across the table from me just staring at their screen going like it's like full of red numbers they said uh are you watching the market it's a bloodbath out there I was like no because I don't actually have any money so it's really unfortunate that your 20 million dollar nest egg is now 16 I guess that's that's really rough for you and they're like yeah you know why don't you come back after the inauguration and see if the world is still here and I was like I literally have no idea how I'm going to make it the next five or six months right so after right around the time we realized we had to put this thing to bed I was helping a friend Dennis Crowley work on the pitch deck for Foursquare back in 2009 and it it came down to first round capital and Union Square Ventures and I had sort of unintentionally touched off a whole VC race to fund this company. I wound up blogging about it. And then Fred Wilson, who I previously had worked for, picked up on what I had wrote, and then he wrote something about it. And then M.G. Siegler, who who was then writing at TechCrunch saw my post and then saw Fred's post and then two dots make a line. So Foursquare must be the hottest company trending out of New York so far. And then all the VCs start swarming like, hey, New York is still here after 2008. That's cool. What's this mobile social app that's the next big thing? So everyone is like really excited about this company. And when the dust settles, USV wound up funding it. And first round had reached out to me and they said, hey, We add a lot of value to our companies. We have some really smart people here, but we feel like we're not as plugged into the New York scene as we kind of need to be. Like, how do we do that? And I was like, you need a person on the ground here. And New York's kind of a funny ecosystem. It's not like the Bay Area. You can't just like swing by Y Combinator Demo Day and meet 80% of the people you need to meet. There's no sort of inner circle in New York. For as much as you think you're part of an inner circle, Maybe you're in the inner circle of prop tech, but you're not in the inner circle of ad tech. And then you're not in the Cornell runway postdoc program or the network surrounding Betaworks or what have you, right? The New York tech scene is very much embedded in various sectors. There isn't sort of a center of the community. And so you just need to kind of go house to house here. And in those conversations, they were like, okay, sounds good. Do you want to come and hang out for a year or two? We don't really have an opening, but we could use somebody doing the footwork. And so I was like, great, because I'm going to max out on my next stage. credit card bill and uh I need a full-time paying balance transfer so this sounds good yeah yeah I'm tapped out and so uh so it worked out the timing worked out I was there for a couple years I did eight or nine deals while I was there in that period of just two years two of them exited so GroupMe sold after a year and single platform sold after two had a bunch of workups like felt like I was shooting fish in a barrel I mean I felt really smart

Brian Bell (00:22:12): based on those financial crisis is almost like the best time to be an early stage investor right I mean I experienced this because I kind of got into the industry in 2020-21 so I saw the big run-up and then I I closed my first fund in 22 so I'm as I'm deploying all of a sudden it's a great time to be an early stage investor you

Charlie O’Donnell (00:22:29): know what that first round of group me was done at 2.83 2.2 2 on 2.8 No, it was $865,000 on a 2.8 pre. That's crazy.

Brian Bell (00:22:43): Priced, right?

Charlie O’Donnell (00:22:45): Were they no good price? I don't remember, actually. Yeah, I don't remember. I have to go back and... 2.8 pre, yeah. 2.8 pre. So at an outcome of like 70... It's actually a pretty decent return. I mean, small dollars and it wasn't WhatsApp.

Brian Bell (00:23:01): What do you think of price sensitivity versus insensitivity in early stage? I mean, this is like a really good example. Like I meet some early stage VCs, they're very price sensitive. I don't go above five caps, which kind of shuts out a lot of the high quality market. And my stance has always been, I'm just a price taker I'm a small check break you just tell me the price whatever and I'll you know maybe get an MFN or something like that and but yeah where do you

Charlie O’Donnell (00:23:28): come down on the sensitivity argument I could there's so many examples where like when I was at Union Square we had a whole conversation about whether or not Indeed was too expensive and Indeed was like say it was like like the founders wanted four on 12 and we had not planned on going above 10 or something like you know like that and and splitting hairs I mean you're just splitting hairs over 20 you know yeah but you know what though like you didn't really imagine at that time that you could sort of guarantee the billion dollar

Brian Bell (00:24:03): outcome and so they were much rarer back then

Charlie O’Donnell (00:24:07): They were much rarer and, you know, so you have to draw the line somewhere and, you know, and that was a company that did not raise after that round. So when it was a billion dollar outcome, it was actually a really phenomenal outcome because there wasn't in between, you know, that outcome. So the problem is, is that you imagine missing out. It's the positive skew of the big outcomes, right? Like if you were an anthropic investor, almost at any price that first round, you would be generationally wealthy by being in that first round.

Brian Bell (00:24:41): Yeah, you'd 100x your fund, right? That's what I like to say. But it put a 25k check in anthropics worth a trillion dollars today. worth billions of dollars after dilution I mean it's an insane return

Charlie O’Donnell (00:24:53): Yeah, but then does that mean you should just go pay $100 million pre for every pre-seed or seed stage deal you do? Like that's not right.

Brian Bell (00:25:02): Well, they have a C in front of their title and they leave one of the AI labs maybe, right? Yeah. Maybe it's a billion. Maybe you're investing at a billion pre, you know? I think of it this way.

Charlie O’Donnell (00:25:13): If someone were to pitch you a fund based around this decision, would you invest in it? And so I go back to like, you know, the Instagram round the Josh Kushner Thrive Capital 2x in 4 days or whatever it was the $500 million valuation that flipped 4 days later for a billion dollars which got him a whole lot of press and, you know, hey, two X in four days is, you know, that's loan shark returns, right? So that turned out pretty well. But if I turn around and I said, I'm pitching a fund and this fund only invests in companies with no revenue at $500 million valuations that have only one logical acquirer and we're going to do that 30 times across the fund. Do you think that would be a good fund to invest in? like probably not you have to be very lucky yeah right like that's not a decision that you could do 30 times and so that's kind of where I think the way I look at this is can you defend that decision over and over again like if if you're like no then your default should probably be no and you you probably shouldn't be in the market now that doesn't mean you can't stray from it but if you have a decision that you can't necessarily replicate To go back to that example of if you were a C-level person at a company that's probably going to IPO at a trillion dollars, well, you're just not going to get 30 opportunities to back somebody like that, right?

Brian Bell (00:26:50): If you're lucky, maybe one in the fund is like that, right?

Charlie O’Donnell (00:26:55): Right and also by the way you're in preferred so that changes the nature too so we're we're quibbling about price but that's a fully diluted price you are also the first money out right so you have a floor downsides kind of protect it a little bit

Brian Bell (00:27:10): this is Here's something that I want to tease that out that a lot of people don't understand that I've noticed I've made 300 and almost 50 investments now over like a seven eight year period is even even when startups they wind down they shut down and they sell for assets basically you normally kind of get your money back you know maybe 50 cents on the dollar maybe maybe a dollar on the dollar but because you're preferred you're kind of first in line unless they have a bunch of venture debt which I won't get into but right

Charlie O’Donnell (00:27:36): It depends, especially early. Before they raise something big, there's aqua hires and maybe they build something useful that they flip to somebody. I mean, I had a company actually reach out to me that was sort of about to go under and then somebody randomly at Staples reached out to me and they're like oh hey this is like a cool app for you know creating some interesting cards and you know it was sort of a little bit of like an it was it was a version of Canva before Canva and Staples reached out and we were basically on a like last few weeks of money and I responded back I was like listen Like this team is going to this opportunity is going to disappear like you guys need to get on your horse and and they came back and they wound up paying 3 million bucks for the assets of the company and the team and everybody made their money back and the founder put a little money in their their pocket and they closed the whole deal in like less than 30 days.

Brian Bell (00:28:34): There you go. Yeah. Magic can happen. So what happened after first run?

Charlie O’Donnell (00:28:38): So there was no real room at the firm, right? These firms don't grow. Some of them don't grow firm forever. Don't grow forever.

Brian Bell (00:28:46): Yeah. I'm like on fun three and I haven't grown. You know, it's really hard to grow at venture. Right.

Charlie O’Donnell (00:28:51): So if you picked up someone, if you had some junior person working with you, there's probably no partnership opportunity. And if you're not a partner, you basically don't have a career in venture. And so... You know, I had a decent track record and I poked around for some other people's funds. I was like, I don't love the idea of going to work for somebody else again and, you know, where my upside is basically capped. Like, you go and work for somebody whose name is literally on the door. It's never going to be your fund. And so... I had a network I reached out and I was like I think I'm gonna do this on my own set up a small fund reached out to a ton of folks and opened up Brooklyn Bridge Ventures the premise being that you know I was gonna do a bunch of super early deals in the New York market where I was like exceptionally plugged in and that is a solo GP I can move faster and go earlier and because it was a small fund I didn't have to sharp elbow anybody out. You give me 30 shots at running into one of the interesting companies and coming out of New York and I have an opportunity to get them. Successfully raised a first fund. I actually raised three funds. Did 105 investments from 2012 to 2023 or four. What was my last deal? About two years ago. Now I'm just working on liquidating that portfolio. It was a fun ride and it was lots of really interesting portfolio companies. It was a very different time period. I think about now what's going on. It feels like a completely different time period to invest than what people are doing right

Brian Bell (00:30:22): Yeah, yeah, especially it sounds like you were deploying Fund 3 during kind of like this run up and run down, right? Is Fund 4 still maybe out there for you? Or you're sort of happy to say, hey, I did my three funds. I've done the run. You know, three funds is a lot of work. Like, you know, hats off.

Charlie O’Donnell (00:30:39): And I'm going to be in this Fund 3 for a long time.

Brian Bell (00:30:42): and that's what people don't realize is like you know I'm still managing stuff I made investments I made five years ago and it really runs 10-15 years right so for

Charlie O’Donnell (00:30:52): sure for sure the big difference especially being a first check investor is now that I'm not screening new deals I have a very different relationship with work and so you know if I were actively screening stuff you might catch me poking around this browser window and Dr. Justin Marchegiani I want to be focused on her. I want my attention to be there. And I think the reality is it's much more competitive now. Everybody's got a little sidecar fund and is a scout and doing all of this sort of stuff. and hyper focused and you know now you have to compete against the 27 year old that you used to be trying to be Johnny on the spot all the time going to five or 10 events a week and being in all the whatsapps and the discords and you know all of And most people who are 46 and do this they either do one of two things they either build a team and they go you know hire the person to do running around or they go a little later right they they narrow the aperture and they say well you know if you don't have a million dollars in revenue or you don't have working product or whatever. I'm only going to look at a certain narrower portion of the window. And that's just not what I wanted to do, actually. I like being the first check, being the first thing to stuff. And I don't love being on a team, to be honest. I don't love having to get approvals and collaborate. I like going at my own pace. And so I was like, okay, I won't enjoy the way I would need to do this now. So I'm just going to go find something else to

Brian Bell (00:32:38): I love that and so that is you know becoming an author and coaching so maybe you

Charlie O’Donnell (00:32:42): can tell us about that yeah sure so one of the things I realized is that the job of a VC for every hundred people in a room is to go find the one or two people you should spend all of your time with I don't love forgetting about the under 98 people you know they they're struggling to raise and a lot of them Some are smart and hustling and have a kernel of something but don't know how to pitch it and aren't familiar with how venture funds work because why would anyone be familiar with how venture funds work if they come out of an industry or they come out of a university and a lab or what have you and I felt like I had something to offer them and so that's what founder unfriendly is. This is basically written for the 99% of people who didn't just exit a company for $500 million and have VCs lining up for them who just pitched or they're about to pitch a bunch of VCs and they're getting, that's really interesting. And then nothing, trying to figure out like, am I, are they actually interested or not? Or, you know, why are they not getting back to me? Or what does this mean? Or I pitched 10 VCs, I got 10 different pieces of feedback and they're all conflicting against each other, like how to make sense of this.

Brian Bell (00:33:54): Yeah.

Charlie O’Donnell (00:33:54): Right.

Brian Bell (00:33:55): And so- Too big of a market, too small of a market. Right. Exactly. Too much traction, too little traction. You know, like literally, you know, too much founder market fit, too little founder market fit, you know.

Charlie O’Donnell (00:34:05): Come back with revenues. Oh, these revenues are service revenues. These are the wrong revenues. Like that sort of thing.

Brian Bell (00:34:10): Yeah. And, you know, it's like... If you've met a VC, you've met a VC, right? If you've met a single VC, you've met a single VC. And so every VC has a different lens that they look at markets. But what's the core claim in the book? One of the core claims is great companies get passed while mediocre ones get funded. So it's not that VCs are dumb, but what's happening there? Some of the good companies are getting passed and some of the bad companies are getting funded.

Charlie O’Donnell (00:34:35): Well, so there's a few things I want to jump into about that is that there are really great business ideas that are not a fit for venture, right? A company that throws off $5 million of free cash flow a year and grows moderately is not going to return NEA's fund. And so especially if it's Dr. Justin Marchegiani if I were a founder, right? So there are some concepts that just do not fit venture as a financial product. And so there's a whole bunch of folks who look at the turndown of VCs and they think that VCs are saying like, this is not a good business idea when it is, in fact, a very solid business idea that is worth owning and worth continuing, but does not have the risk return profile for this kind of financial platform.

Brian Bell (00:35:32): Yeah, it's kind of like if it works, it's not going to be a $10 billion company. If it works, it's going to be a $50 million company. And if I'm getting into the company at a 5 or 10 cap valuation, it just does not pencil with my model.

Charlie O’Donnell (00:35:45): Right, exactly. But if you're an individual angel investor or if you're putting your own capital in, that might be a really nice outcome for you. And so that's one set of companies. There's another where the underlying idea is quite good, but you have just not successfully indicated to VCs that it is your intention to grow a big company or that that possibility exists. And so so many founders kind of come to VCs and they say, I have this. What do you think? Is this good? And that's essentially what their pitch sounds like Versus saying, my intention is to drive to a company that is going to do $500 million in revenue Because if you come up to a VC and you're like, I have this, is this good? And then there's somebody else who says, I am staring at an opportunity to drive a company to $500 million of revenue in eight years That could actually be across some aspects A worse idea, but you have just indicated to the VC that like, here is my intention, I'm quite capable. A VC is not going to assume that is your intention to drive something to a $500 million idea, especially because doing that at that trajectory also increases the downside. It increases the chance of a zero. So some founders unintentionally signal that they don't want to take those risks. So for example, I had a fairly diverse portfolio of founders. And one thing I see for founders that don't look like me, whether it's female founders, founders of color, is that thank you society, but we are sort of programmed to imagine that anybody who doesn't look like me is less aggressive, is less ambitious, right? And so if you go in and you're a female founder and you present a roadmap that like A, either only has the next two years of financials because that's the round you're raising now and that's where you feel confident about where you're going to spend the money and all of this sort of stuff and like literally leaves the rest of it open or has the forward trajectory but it is quite conservative because your perception is I'm under a lot more scrutiny than my male counterparts so I don't want to screw this up right I don't want to say that I'm going to get to 200 million dollars revenue and then have it only come in at 150 and then you know I lose my job as CEO I placed or all this sort of stuff Okay, I get it. Except that you are literally walking right into the existing bias that they have, right? They assume you're not as aggressive. And then when you're projecting something that you believe you can promise versus what the potential is. So I always used to flip the conversation and I would go back to the founder and be like, I see you're raising a million dollars. Would you really not know what you would do with $2 million? And I'd have those founders come back and she would say, of course I know what I would do two million dollars I would double down on product here I'd hire twice as many salespeople I'd open up in three cities instead of one well and what do you think that would do to your revenue oh that would really open revenue up like we would easily get to twice the revenue in half the time or whatever cool that pitch is way better than what you just gave me Why didn't you pitch that? Well, I can't promise that. Right. Nobody can promise anything. I can't promise that I'm not going to walk out of my apartment and a piano is going to fall from the sky. Like I can't promise that. Right. But this is not a promise conversation. This is a potential conversation because you're sitting across the table from a VC who knows that only one or two of the 30 deals they do in this fund are going to return the fund. They don't know which ones, but they need to make sure that they all have that potential. And if I don't hear it from you, then I'm not going to assume it, right? Because I don't know that that's your intention to drive it. I don't know that's what kind of trajectory. I don't know that's how much you're going to put your foot on the gas. And so... There are ways in which founders sandbag the pitch unintentionally, and they need to understand what's going on in that room, what's being asked of you. What's being asked of you is to have an exercise that identifies the potential of this thing, not to promise what you know will happen.

Brian Bell (00:39:56): So it sounds like what founders are doing unintentionally is thinking small, right? They're thinking small.

Charlie O’Donnell (00:40:03): They might be thinking big, but they're afraid to say it.

Brian Bell (00:40:05): Yeah, and VCs need to hear a big vision and a big plan, right? And this dial of money does different things to the outcome of the company. So that's really good advice. Tidbits and stories can you draw out of the book for the audience?

Charlie O’Donnell (00:40:21): I feel like you learn a lot about pitching when you're the parent of a small child who has the attention span of a gnat, right? Because it's like, go in your room and put your pants on. And then all of a sudden you hear, you know, the Yodo player going or whatever. It's like, pants, pants, just focus on just the pants, right? Like, think of pants as fund returning outcome, right? Every slide needs to scream pants because we'll just kind of get lost, right? VCs have no attention span. There's not an IQ licensure. Just because they're good at collecting money from institutional investors or whatever doesn't necessarily mean they're the smartest people in the room. It's usually the founders. And so from the beginning of the deck, the outcome needs to scream potential fund returning outcomes. And I think there's a couple of reasons why that happens, right? Certain groups of founders feel they need to justify their presence in the room. Let me tell you how I got here. I was hired by this person. That's impressive. I went to this school. That's impressive. Here's how I even got into knowing about this because you might question how I'm an expert in this even though my background doesn't show up. Five slides in, I don't know if you've made any money I don't know if this thing's even live I have no idea how many signups like you're busy telling me your backstory and it feels like you know back in Sicily in 1922 kind of you know a story right so revenue Traction whatever it is the other thing and I just you know jumped into showing revenue and immediately people listening are like what if I don't have revenue yet right there's too many templates too many shortcuts people are not good at figuring out the interestingness of their own story and here's the way I think of it You basically have team, market, and traction. I'm not going to say product because VCs are actually really bad at figuring out product, right? So they'll tell you like, oh, I have this thing and it's a box and it's blue. And it's like, I guess that I'm not, I don't need boxes. So I don't know whether or not that's the right kind of box for this customer. But traction is a proxy for product. I wouldn't have used that, but I guess people are buying that, right? I guess people are engaging. you have 10 design partners that's crazy like why are people even engaging with you on this at this stage this must be a real problem right usually at the sort of pre-seed and even the seed stage you kind of have two out of the three like you're very rarely going to have all three and so your particular narrative might be team and market like we've assembled this incredibly Like experienced group of people and we're going after this space that you definitely want to be a part of. Now we haven't actually done anything yet, but don't worry about that because this is the bet. You're going to want to make a bet on us in this space. Or we're awesome and this is a train leaving the station. Forget about what you've heard about this space and everything that everybody's told you that this space sucks. The lead of this conversation is the two out of the three, team and market. Or This is an amazing market and the train's leaving the station. Yeah, we're a bunch of nobodies, but that doesn't matter, right? Because we've got this two out of the three, the lead. And so you need to craft your narrative so that the most impressive, the most important stuff is up front.

Brian Bell (00:43:47): And everything else is kind of played down and you keep that energy level,

Charlie O’Donnell (00:43:50): you keep that excitement. And you need to spend most of that money talking and most of that meeting talking about the upside. And this is where a lot of people get tripped up because you may have seen the Harvard Business Review study that women get different questions. They get the downside focus questions. We've all seen that, right? And they say, well, how do I even get around that if they're asking me bad questions? Have you ever seen a politician speak in an interview? Do they ever answer the question that they were asked? They have their talking points. I'm so glad you asked me about jobs and the economy because what it really comes down to is making sure that That's why my crime bill is what the fuck just happened? We were trying to make it an economic conversation. We immediately pivoted to putting more cops on the street, right? Because if you can't get to work, if you get mugged on the way to work, then you're not going to be able, the economy is not going to be able to function, right? So you need to make sure that you spend the bulk of the time in the meeting on the upside case, on the things that you wanted to talk about, on your message, regardless of what questions are happening in the meeting. Too often I see founders just let go of the reins and just let VCs control the meeting. And that doesn't work out.

Brian Bell (00:45:00): Yeah, I love that. And it's actually a really good example of kind of being in the room with a founder that is just a force where you're like, okay, this person has that executive presence. And I think part of being a good VC is being able to kind of flesh that out a little bit, right? Where you're like, okay, this person is going to, you know, climb mountains and tear down walls and cross rivers and they're going to get people to follow them, right? And that's a little bit of the, you know, the art of this business, I think, right? Because you can look at traction objectively, you can look at, you know, metrics and even the product, right? I'm a product guy, but a lot of products, I look at the product and, but really, it's kind of like, do you see this executive presence that's going to get, you know, operators and execs from other companies where they make, you know, two to three X what they'd make at the startup, at least in the short term, to join them and risk their livelihoods and careers to work on this crazy idea, right? So it's a little bit of that, you know, you might call that Steve Jobs kind of factor, right? Reality distortion field or

Charlie O’Donnell (00:45:57): All of those things, right? Because you're signaling, somebody's sitting there imagining like there's a killer CMO that we need to hire and we have no business hiring them. Can this person convince them to come and work for them, right? We need the New York Times to go use our payment platform and nobody there wants to switch anything out there, you know.

Brian Bell (00:46:19): Even though they're maybe the most cutting-edge digital media company,

Charlie O’Donnell (00:46:23): they're still a 200-year-old media company. How are you going to do that? And if you come in and you let VCs run the meeting, then they're not going to imagine that you're going to do those kind of impossible things that it takes to get two people up to an IT.

Brian Bell (00:46:40): So take control of the meeting. Be that annoying executive that just commands the room, basically.

Charlie O’Donnell (00:46:46): I think there's ways to do it without being obnoxious I think there's one of the things I talk about in the book is some structure right just a very polite useful structure hey I'm gonna spend the first five minutes just making sure we're fit and I'll tell you a little bit about the company and you tell me whether or not this is a good sweet spot for you or whether or not we're just gonna like stay in touch or whatever and then you go in you say well there's the people who've gotten excited about us Get excited about these three things, right? And it's A, B, and C. If I convinced you of those three things, do you think you'd be going back to your team and championing the team spending more time with us, right? Or is there something else? Like is there, you know, have you looked at this space and you think it's not three things, it's these three other things? You let me know what you're looking for in this space. cool and then at the end of the meeting you circle back hey I tried to convince you of these three things this is what we spent our time doing you opted into it in the beginning of the meeting right this meeting had some structure do you think I did a good job here right like so it doesn't have to be a rudeness or overbearing it could just be a structured like you've bought into a certain dynamic of an exchange here and so I created spaces for you to ask certain questions about what I was trying to convince you of and guided the nature of the conversation it was a polite back and forth exchange and I think you can listen anybody who has a sales training background these conversations are Structure, they're intended to get information back from each person, right? And done well, it feels like I'm along for the ride, but I'm not handcuffed, right? Like I'm not, I've been kidnapped into this. I've been gently guided into an exchange that makes sense and feels productive. That's great.

Brian Bell (00:48:38): Well, for any founders listening, you should definitely go read that book. And, you know, I always, you know, know it's a great conversation when we can't even get through the outline. Like there's just so many questions. I tend to be a little I'm like there's just no way we're gonna run out of time so why don't we we'll probably have to have you back on at some point and we'll we'll talk more because has been really informative I think especially for founders listening it's just there's just so much you know in 20 years in the industry that you have all I appreciate that, thank you Let's wrap up with some rapid fire questions If you had to build your fund one company from the last decade that you passed on with your own money today, who and what did you miss and what would you wish you had invested in?

Charlie O’Donnell (00:49:30): So wait, sorry, if I had to redo my fun one with perfect knowledge, I mean, that's an easy one. I did a really good job generating top of funnel and I wish I would have had AI help me process the funnel and identify the higher potential stuff because email is just really tough to manage. And so I have an email from the founders at Plaid that is Hey, we really love your newsletter. We'd love to come and talk to you. And I responded back to them. I used to have this like meeting request tool. So you could like, you know, sort of like before Calendly. So they requested a meeting like tomorrow, but I didn't get it until like right after the meeting. I was like sorry to miss this actually I'm doing an entrepreneur gathering for lunch this Friday like you should join me right and like they didn't catch that response and I didn't follow up and that was it and so like who knows if I would have invested in Plaid or whatever but like the funnel did its job like I got an inbound from Zach that was like love reading your stuff would love to share and then it just kind of fell through the cracks

Brian Bell (00:50:42): I think the network side of it, the fact that if you have five venture-backed friends, you are at a huge advantage. Take a look at my deck, how do I not screw up this co-founder hire?

Charlie O’Donnell (00:51:07): What am I looking for in a VP of sales? The experiential knowledge of being successful on the startup side is not evenly distributed. There are some founders who are not in the center of those networks that will have to spend 10x the amount of time building those relationships versus other people who just lucked into them or privileged into them or schooled into them or whatever. but those people are just better founders because they have that on tap and it's not fair but VCs are just trying to find the best founders and the best founders are the folks that have these networks frankly and so you know you need to double down on making sure you're connected to everybody you need to be connected to yeah

Brian Bell (00:51:51): that's smart so Steelman the founder friendly era you partly wrote against what did Brad Feld and his generation get right that your book undersells

Charlie O’Donnell (00:51:59): what did brad felt and his generation get right that the book undersells i never really thought about it that way i mean when i think of what brad got right is i think what brad gets right is just being a good person in in venture she's like brad's just a really good guy right yeah And there are a lot of people in venture that I would say are not good people. And I would like to believe that success disproportionately rewards good people. I do not think that is the case. However, I think happiness does reward good people. I think it's really hard to be an extractor of value, to be somebody who is dismissive of people not like them, to be constantly on the attack, to be constantly in negative conversations with people people in, you know, a reductionist of people and to ultimately be happy and maintain real friendships and, you know, all of that sort of stuff. So you can be rich and unhappy or you could be rich and happy if you're a good person. So I think I'd rather be slightly less wealthy or not wealthy at all and be happy and a good person.

Brian Bell (00:53:22): Yeah. And it's not always a black and white decision either. You can be rich and happy.

Charlie O’Donnell (00:53:27): I always say that when I think of levels of wealth the way I look at it is one level of wealth is living exactly where I want to live in the kind of apartment that I want to live in and we're like we could use an extra room but that's about it like we're pretty close right and then the level that counts for me above that is majority ownership in a major league baseball team and like I'm not going to get there I don't know what's in between there to be honest I suppose I could buy art but like that's if I yeah if I can't buy the meds from Steve Cohen then I'm fine with

Brian Bell (00:54:02): the amount of money I'm fine flying first class I don't need to fly private you know right that's like my level of wealth is just that's fine that's good for as a poor boy from Seattle that's that's good enough for me there you go what's a sentence or a message you'd love to send to your 2012 version of yourself just launching Brooklyn Bridge BC knowing how it ends now what message would you send back to yourself

Charlie O’Donnell (00:54:23): Don't be dismissive of what you had at first round and because when I was at first round capital and I went to go pitch the number one LP question was that like your track record is this just Josh Koppelman's leftovers or like can you still get good deals not being at first round and I very much think Yeah, yeah and I should have internalized some of the things that I had and say, well, you know, obviously this is different, but, you know, here's the way I am leveraging my own version of what I had before. And so I think like since that time and generally speaking, I try not to be dismissive of constructive criticism.

Brian Bell (00:55:24): Yeah. What's a belief about VC that you held dearly but have now reversed or changed your mind on or vice versa, something that you didn't believe but you now believe?

Charlie O’Donnell (00:55:35): this is an easy one actually about 50% of the companies that I backed did not have product when I first wrote the check and I used to think and it was actually true that there was a price arbitrage that was worth it at the time where there were some complex things where you could get in on the cheap for a good team that you had absolutely certain absolute certainty would get to product right now whether it would scale or whether people want to buy it would be a whole other thing but just like having product I think people disproportionately paid for it in a pre-AI era. I don't think that's not the case anymore. The bar for product is so much lower that somebody who has nothing is now adverse selection in the way that they were not five or ten years.

Brian Bell (00:56:20): Unless it's like deep tech or something like that. But yeah, if this is a software product, you should have something. Yeah, before you're pitching VCs for sure. Well, I really enjoyed the conversation. I could have talked to you for another hour at least. Where could folks find you online and find your book and stuff?

Charlie O’Donnell (00:56:33): So the book is wherever books are sold in whatever format people buy them although we don't have an audiobook yet and I am actively trying to get my publisher Are you

Brian Bell (00:56:42): going to read it yourself or are you going to hire...

Charlie O’Donnell (00:56:44): I would very much like to and we are going to trial some professional readers and I'm sure they're very good but I would very much like to do it myself so we'll see and all of my I continue to write I mean I've been writing my newsletter slash blog since February of 2004 at thisisgoingtobig.com and so I will continue to do that it's actually very hard for me to just send a version of the book out where I can't change it right somebody's a regular newsletter writer that's very uncomfortable yeah so I subscribe to me there I may contradict things that are in the book and I'd be totally fine with doing that because the world keeps changing but that's a good place to find me and my writing and what I'm well thanks so much Charlie really

Brian Bell (00:57:29): Enjoyed it thank you

That is usually the point.

Great design disappears into the experience. It reduces confusion, hides complexity, and makes hard technical systems feel obvious. But behind that simplicity is a long chain of decisions: what gets built, what gets cut, what gets explained, what gets tested, and what actually makes it into the customer’s hands.

Lauren Von Dehsen has spent her career working inside that chain.

Across Nike FuelBand, Nest, Google Health, Matter, and SoFi, Lauren has helped shape products at the intersection of hardware, software, and human behavior. Her work has touched connected devices, smart homes, healthcare, fintech, and now the emerging world of AI-assisted design.

Her biggest lesson is blunt: documentation is not the product. The process is not the product. The customer only judges what ships.

That idea sounds obvious until you watch how teams actually work.

Subscribe now

The Product Is the Only Thing the Customer Sees

Early in her career, Lauren wrote about the idea of designing the product, not the documentation. At the time, she was working in environments where hardware and software had to come together in new ways. These were not mature product categories with well-worn patterns and obvious playbooks. Teams were inventing the product, the process, and sometimes the testing methods at the same time.

On Nike FuelBand, for example, people were literally running up and down stairwells to test whether the device was tracking activity correctly. That is the reality of building something new. You do not always have a clean lab environment, a perfect spec, or a known answer. You have a product that needs to work in the messiness of the real world.

For Lauren, that shaped a core belief: a designer cannot just hand off polished screens and assume the job is done. The real work is making sure the idea survives contact with engineering, testing, constraints, trade-offs, bugs, edge cases, and launch pressure.

A beautiful design file does not matter if the shipped experience fails.

That mindset is especially relevant for startups. Founders often confuse design with surface area: colors, screens, typography, polish. But design is more fundamental than that. It is the practice of making decisions about how a product behaves, how people understand it, and how the experience holds together under pressure.

Nest and the Power of Cross-Functional Design Culture

One of the chapters Lauren looks back on most fondly is her time at Nest.

Nest was not just a company that made better-looking thermostats and smoke detectors. It helped redefine how people thought about everyday objects in the home. Before Nest, most people did not spend much time thinking about their thermostat. It was a beige plastic box on the wall. Functional, forgettable, and usually ugly.

Nest changed that by bringing software-level interaction design, hardware craft, and brand-level attention to a category that had been ignored.

But the real lesson from Nest was cultural.

Lauren described an environment where she did not have to constantly explain why design mattered. Everyone cared about the product experience. Engineers, product leaders, designers, and other functions could debate almost any aspect of the product. The difference was that, after the debate, the team trusted the person closest to the decision to make the final call.

That balance is rare.

In weaker cultures, design is either isolated as “the designers’ job” or diluted into endless committee feedback. At Nest, design was everyone’s responsibility, but designers were still trusted as experts. That is a powerful distinction.

For founders, this is one of the most important takeaways from the conversation. If design matters to your company, it cannot be something you bolt on at the end. It has to be part of how decisions are made from the beginning.

What Founders Get Wrong About Design

Early-stage companies often bring designers in too late.

A founder may define the product direction, product managers may shape the requirements, engineers may begin scoping the system, and only then does someone ask design to “make it usable” or “make it look good.”

That is backwards.

Lauren argues that designers are most valuable when they get context early. Not when every decision has already been boxed in. Not when the team simply needs a screen. Early context allows designers to understand the customer, the business goal, the constraints, the technical trade-offs, and the hidden assumptions behind the product.

That does not mean every startup needs a huge design team. It does mean founders need to be honest about what they expect design to do.

Are you hiring someone to make the MVP presentable?

Are you hiring someone to define the customer experience?

Are you hiring someone to help shape product strategy?

Those are different jobs. They require different levels of seniority, different skill sets, and different levels of organizational trust.

The mistake is not choosing one path over another. The mistake is pretending you want strategic design while treating designers like production support.

Scaling Design at SoFi

Lauren joined SoFi in early 2020, just weeks before the pandemic changed how teams worked. At the time, SoFi was in a high-growth phase, operating across a wide range of financial products: banking, investing, lending, crypto, and insurance.

That created a very different design challenge from Nest.

At Nest, the work centered on tightly integrated hardware and software experiences. At SoFi, the challenge was scale, complexity, and coherence. How do you create one unified customer experience across financial products that behave very differently? What should be shared across the brand? What needs to be specific to each vertical? How do you build a mature design and research organization that can keep pace with a growing company?

Lauren eventually scaled SoFi’s design and research function into a 100-person organization. That required more than hiring. It required building rituals, processes, expectations, and cross-functional relationships that could evolve as the company changed.

One of her points is especially useful for leaders: rituals have to serve the outcome. They cannot become the outcome.

Design critiques, reviews, sprints, and research processes all have value. But as companies scale, the same rituals that once created alignment can become bottlenecks. A critique with five people may be useful. A critique with 30 people may be theater.

Good design leadership means knowing when to change the system.

Process Is Useful Until It Becomes the Point

Design teams love process. Design sprints. Double diamonds. Workshops. Critiques. Frameworks. Research panels. Naming conventions.

Some of that is useful. Some of it becomes internal language that does not help the broader company.

Lauren’s view is practical: use the tool if it helps the team get to the next decision. Do not worship the tool. Do not over-explain the ritual. Do not assume cross-functional partners care about the purity of the method.

Most partners want to know what decision is being made, when the answer will be ready, and whether it will help the product move forward.

That does not mean design should become reactive or shallow. It means design leaders need to translate their work into business-relevant outcomes. The best design process is the one that helps the team build a better product faster, with fewer blind spots.

Anything else is overhead.

AI and the Next Wave of Design Tools

The conversation also turned to AI and how it is changing design.

Lauren sees clear productivity gains. Designers, like everyone else, can use AI to brainstorm, write, summarize, explore concepts, and accelerate early work. But she also sees a major limitation: design is not only language.

For many software tasks, moving from keyboard input to conversational prompting is a relatively natural abstraction. But design often involves spatial judgment, visual hierarchy, motion, color, breathing room, sequencing, and subtle interaction details. Describing those details in words can become frustrating fast.

AI tools may generate strong first concepts. But as the designer tries to refine the work, make precise changes, and bring the output closer to a specific vision, the process can become fatiguing. The more exact the desired change, the harder language-only prompting becomes.

This is why Lauren is interested in tools that combine chat-based interaction with direct visual manipulation. The future of AI design probably will not be pure prompting. It will be a hybrid interface where designers can generate, edit, manipulate, critique, and refine in the same environment.

The open question is which tool becomes the design equivalent of the AI coding copilot.

The Hardest Design Trade-Off: Craft vs. Reality

One of the most honest parts of the conversation was Lauren’s reflection on how her own thinking has changed.

Earlier in a design career, it is natural to want everything to be elegant, complete, polished, and deeply considered. That instinct is valuable. It creates standards. It pushes teams beyond mediocrity.

But leadership requires a different kind of judgment.

Sometimes speed matters more. Sometimes a business constraint matters more. Sometimes the right call for the company is not the designer’s ideal recommendation. Lauren described moments as a leader when she had to make decisions against her team’s design preference, not because the team was wrong, but because other constraints had to win.

That is the maturity curve of design leadership.

The goal is not to abandon craft. The goal is to know when craft is the decisive variable and when it is not.

The Real Job of Design

The through-line in Lauren’s career is not just design excellence. It is systems thinking.

FuelBand required understanding the relationship between a device, an app, and real human movement. Nest required turning overlooked household objects into intelligent, trusted products. Matter required thinking about how devices communicate across ecosystems. Google Health required building consumer UX in a deeply sensitive domain. SoFi required scaling design across a complex financial platform.

In every case, design was not decoration. It was the connective tissue between technology, behavior, business, and trust.

That is the real lesson for founders and product leaders.

Design is not what happens after strategy. Design is one way strategy becomes real.

It is how a product explains itself. It is how complexity becomes usable. It is how teams make decisions visible. And, ultimately, it is what customers experience when all the internal debates, documents, meetings, trade-offs, and intentions disappear.

The customer never sees the process. They only see the product.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:
00:01 — Introducing Lauren Von Dehsen
00:48 — Lauren’s Origin Story in Design
01:25 — Choosing Design Over Math and Science
02:56 — Discovering the Design of Everyday Things
04:31 — The Product Lauren Is Most Proud Of
05:43 — Joining Nest During the Google Acquisition
06:48 — Why Nest Changed the Smart Home Market
08:11 — The Timing Behind Nest’s Breakthrough
10:28 — Design the Product, Not the Documentation
11:21 — Why Great Concepts Often Never Ship
13:14 — Taking Ownership of What Customers Actually Use
13:40 — Designing Across Hardware and Software
15:09 — Inside the Secret Nike FuelBand Project
16:27 — Asking the Questions Nobody Had Answered
17:21 — Designing Before and After Figma
18:01 — From InDesign Specs to Collaborative Design Tools
19:56 — How Figma Accelerated Product Design
21:16 — How AI Is Changing the Design Landscape
22:19 — Why Prompting Is Harder for Visual Work
24:15 — Claude Design, Noon, and the Future of AI Design Tools
25:20 — Why Design Needs More Than Words
26:13 — Lauren’s SoFi Chapter
26:30 — Leaving Google for a Faster-Stage Company
28:03 — Why New Problem Spaces Create Better Design Thinking
29:34 — Joining SoFi Right Before COVID
30:27 — Building a Mature Design and Research Organization
31:11 — Designing Across Banking, Loans, Investing, Crypto, and Insurance
32:04 — What Founders Get Wrong About Design
33:19 — When to Use Familiar Patterns vs. Diverge
34:26 — What Nest Got Right About Design Culture
35:21 — Trusting Designers to Make the Final Call
36:24 — Moving Design From Execution to Strategy
37:54 — Applying “Product, Not Documentation” to Design Leadership
38:21 — Rituals, Reviews, and Scaling Design Teams
40:10 — Thread, Weave, Matter, and Smart Home Interoperability
41:23 — Designing for Devices That Need to Work Together
42:17 — The Most Common Early-Stage Design Mistake
43:32 — Bringing Designers Into the Conversation Earlier

Transcript

Brian Bell (00:00:56):

Hey everyone, welcome back to the Ignite podcast. Today we’re thrilled to have Lauren Von Dyson on the program. She spent 15 years at the intersection of hardware, software, and human behavior, designing products that people interact with every single day without thinking about it. She shaped the UX of the Nike Plus Fuel Band, very cool, when wearables were still a bet. Helped define how smart home devices talk to each other at Nest, very cool. And then informed the Matter Protocol now adopted by Google, Apple, and Amazon. Built Google House Consumer UX Org from scratch. and then spent six years scaling SoFi to a 100-person design and research operation. And the list goes on. Thank you so much for coming on, Lauren.

Lauren Von Dehsen (00:01:34):

Absolutely. Thank you so much for having me.

Brian Bell (00:01:36):

Well, I’d love to start with your origin story. What’s your background?

Lauren Von Dehsen (00:01:39):

I come from an industrial design background academically. I grew up in New York. Maybe that’s also helpful to hear. And I went pretty quickly into technology and software development following graduation. So it was a great entry point into the world to kind of know that I wanted to pursue design much earlier than I find some of my colleagues knew that and getting kind of a head start into the space.

Brian Bell (00:02:01):

Yeah. So what was the aha moment where you’re like, I’m doing this. I’m doing the design thing.

Lauren Von Dehsen (00:02:06):

Well, I’ve been credited as a very deep and sometimes overly deep thinker. is maybe the best way to say it and going into my college search I was coming from a background academically that was very strong in math and science both of my parents worked at IBM and were in various forms of engineering and project management but I had this kind of lucky extra context that my aunt and uncle had gone to Parsons and were both graphic and industrial designers and had been pursuing that career and I grew up around that and got to see what they were doing so I’m kind of a product of my environment and circumstances but when I was looking through colleges and trying to really be thoughtful about what I was going to pursue I found that as much as I loved the academics of math and science and learning it and studying it I couldn’t see myself being as excited about the careers I understood that would lead to which wasn’t necessarily all of the opportunities but the ones that I had on my radar And when I thought about the times where I was doing creative pursuits, whether that was alongside my academics or during the summer, I found that I had a different level of passion and interest. And so I’m very fortunate that it kind of all came back together and I actually used both sides of that. But I didn’t know that when I was pursuing it and took the leap of what felt like a very risky leap to go down the design path.

Brian Bell (00:03:25):

Yeah and I didn’t really even get exposed to design until I became a product manager right and I came into product management from marketing and so I was even maybe in marketing a little bit with landing pages and things like that and email layouts, but really it was like product management. I was like, oh, there’s this whole like world of how things are made. You know, I think one of the first design books ever was Design of Everyday Things. Yes. It’s just such a great book. Classic. How are doors, you know, like doors designed, you know, like you could have a knob and you can have a plate and you can have a bar and like what are the different reasons for those different affordances, right? To open and close doors and you just learn this kind of stuff.

Lauren Von Dehsen (00:04:04):

I think it’s a real eye opener at least for me I think I had this moment and I can’t quite place maybe what age it was at but I had this moment where I went wait a minute everything around us somebody made a decision about whether it was an intentional decision whether it was I’m what’s a product that you’ve worked on that you’re just really proud of that you look back and you’re like that was my best work I have a soft spot for anything that that I put hours into, right? Because like anyone that works in the tech space, you know, you really invest a lot in making these products. I think the one that always, the chapter that always stood out the most to me was the Nest product suite because they felt so substantial and monumental and there’s a level of excellence that the team and the company really committed to, particularly in the early years and the era.

Brian Bell (00:05:27):

Before Google acquired them, yeah.

Lauren Von Dehsen (00:05:29):

Well, I joined right in the midst of the acquisition. There was a pretty big hiring spree, right? Because with the backing of Google, there was means to do it. But I like to say, I think when I joined, I used to think, how do I make people think that I was here before acquisition. I want to know all the, you know, the details and make sure that I know this as best as I can. You know, I think that was out of just a love and adoration for what the initial early team had done and wanting to continue that forward as we expanded into what, you know, was kind of the height of the product suite that Nest put out across HVAC, smoke detectors, cameras, security, doorbells, etc.

Brian Bell (00:06:07):

And so... Talk about the design of everyday things. I mean, you know... Very everyday You know smoke detectors

Lauren Von Dehsen (00:06:15):

the things that I think before that era, you wouldn’t even have really spent much time or energy thinking about. And now they’re, you know, these somewhat high tech tech devices. And I think there was just such a beauty and excellence and attention to detail that the Nest team was able to deliver. And it was something that was just a standard that was held across every function, everyone on the company, no matter what they were contributing to the success of it. And so when I think of those products, I just think of them so fondly because um it just felt like the team was firing on all cylinders yeah and everybody knows

Brian Bell (00:06:49):

and loves Nest products if you know it’s a great interview question actually you can you know ask a product manager or designer you know tell me about your favorite product right and why you love that product and a lot of people love their Nest products what do you think it was about kind of the timing of of Nest I think about this a lot as a VC we care about timing a lot you know like I wonder like what year was it founded like 2009 or 11 somewhere in there yeah you have the right hour I go somewhere in there you know and I often think about this as a VC like if the Nest founders you know sent me their pitch deck or you know they would probably get an intro hopefully get an intro would I invest right around you know what do you think it was like looking back because up until that point we were used to kind of you know the oddly shaped plastic box thermostat that was just what it was is the Honeywell or whatever it was and that’s just how they worked it’s like they had three ugly gray buttons and it was like this yellowing box that hadn’t changed that much but we got a little LCD display you know whereas before it was like you know the stuff that we grew up with in the 80s or whatever what you know what do you think it was about the timing where like Nest could succeed when it did how it did

Lauren Von Dehsen (00:07:59):

I’ll give you my perspective obviously there are some that I could really tell you the facts in a way that, you know, they firsthand lived it. I think with something like that, you have this kind of perfect storm of components. I think there was a realization that technology, both hardware technology and software technology, were getting to a point that made some of the technical components possible. What I mean by that is, you know, the guts of the first thermostat, the way that I I’ve always thought about it was it was really a more rudimentary cell phone and because we had many generations of the iPhone never mind the fact that the people starting Nest really knew those products intimately you know they could carry that over almost seamlessly into a different I’m phases of technology or themes in the industry. There’s this catalyst of like something made it possible. And then you have a team that’s interested in a particular problem that’s well suited to what the technology is allowing. I think with the early days of Ness, my understanding of it always was that there was such a push on the environmental side because the thermostat in particular had such a connection to how we use energy and power in our homes and that I think struck a chord with conversations that were going on and what was out in the world that people were starting to be wise to and interested in doing their part and so that’s where yeah the technology is the enabler but then you have these other maybe additive pieces that when all happening at the right time give that great environment for something to pop up like nested

Brian Bell (00:09:56):

And I guess everybody had that, you know, perfectly crafted iPhone experience around the same time, too. And you’re just expecting things to, you know, feel great. And the thermostat definitely did not until this came along. You published a piece called Design the Product, Not Documentation. do you recall that paper and like what the core argument was yeah it was a long

Lauren Von Dehsen (00:10:15):

time ago and I was very early on so I think part of my motivation to write was to just organize my own thoughts and learnings and how I was approaching design in the early years of my career but that was particularly an insight I think partly driven from hardware development it might have been written more in the fuel band era than the nest era but same you know same general idea and what I realized was that a lot of designers would be really proud of the concept that they came up with and the thing that they had shared with their team and a really innovative idea but the amount of those that make it to what actually ships as far and few between kind of by nature of what we do. And what I started to realize was that, yes, I needed to make documents and specs and I needed a way to communicate with engineers and team members to express my ideas, but that It kind of didn’t matter what was in that file if it didn’t actually make it into people’s hands. And what it took to make it into people’s hands was explaining myself, you know, 10 times more than I thought was necessary and double checking my logic five times, even though I was pretty convinced it was the right call. There were new pieces of information that kept making me have to rethink a decision that I had made weeks ago or spending hours just you know bug bashing which of course is such a standard thing we all do today but we were you know learning how to do that on mobile apps and in more interesting platform environments that we hadn’t had before and we were learning how to do that across a hardware device and an app that spoke to each other which was kind of a harder thing to wrap our heads well how are we going to test this we had people in the fuel band days we had people running up and down the staircase the emergency staircase in the office building trying to figure out how to get some points on the band so when you’re making something that doesn’t I think the motivation behind that piece was, as a designer, this is my problem too. I can’t just sit on my laurels that I made the screens and I made the flows and I gave it on time so that nobody downstream of me was held up. I have to have ownership in what actually makes it into the customer’s hands because that’s the only thing they’re ever going to judge the product on.

Brian Bell (00:12:35):

Speaking of FuelBand, you were at RGA. Do I have to say Slash?

Lauren Von Dehsen (00:12:39):

No, no, no. RGA is correct.

Brian Bell (00:12:41):

You know, there was this moment you’re working on a connected device. Nobody knows. And this is kind of a recurring theme in your career, I guess, which is kind of this hardware software integration. How down into the industrial design are you going versus staying at the kind of the software layer?

Lauren Von Dehsen (00:12:55):

I always worked with industrial design teams. I never practiced industrial design professionally, even though that was my kind of degree origins. And so the scenario would, it would be different on different projects. If the device had a lot of buttons and direct interaction on it, there might be more reason to collaborate in those early phases versus a device that maybe was more simplistic and we had to just make sure enough of the pieces of hardware were correct that would give us flexibility down the line of the the project lifecycle to determine how the software would use that hardware. So I would say it was pretty sliding scale depending on the type of product and the specifics of what that needed.

Brian Bell (00:13:36):

What was that like working on the fuel band? I mean, that was one of the first connected devices and guessing you had to bring your design the product, not documentation mindset into that.

Lauren Von Dehsen (00:13:45):

Yes. Yeah. I came into it. The project was already kicked off and I joined the team a midway towards the first launch so it was totally secret I was exposed for the first time to like a locked down environment where we have to go through multiple

Brian Bell (00:14:00):

layers of security to get to the devices and things like that that’s right exactly

Lauren Von Dehsen (00:14:05):

it was just it had so much allure around it that I think I was just you know taken by like oh my gosh I can’t believe I’m in here and getting to see how this how this happens and then you know you get into the project and again this is pretty early in my career I might have been like designer senior designer at best at that point you get into the project and you think okay all these like super smart people have worked through all of this and they’ve already been on it for however many weeks if not months and I just need to like do what I can to help and plug holes and then you start running into things like any product development right which you know I was naive to at the time where we didn’t we never talked about how that thing was going to work which just kind of either been overlooked or it’s just working right now as the assumptions that were made along the way as an output of everything else that was decided. And I started to realize that, you know, I was kind of just assuming somebody knew better. And one of the things I think clicked for me was, no, no, no, no, no, just assume nobody had the time to think about that and that it just went under the radar and you’re better off asking the question and finding out there is an answer than or obscuring it or just sitting on it thinking somebody knows this. And so I found one of the things that was really great is that the more that I asked and the more that I poked, the more I realized how new everyone was to it. It didn’t matter their years of experience. The problem space was so new and we didn’t have patterns for it. We didn’t even have some of the technology setups we needed to really be successful. And so you have to kind of design the tools and the process alongside the product itself.

Brian Bell (00:15:41):

So you’ve been a designer, what you might call before Figma and after Figma. How did that change? I remember when that came out and I like what a sea change that was for me as a PM.

Lauren Von Dehsen (00:15:52):

Yes.

Brian Bell (00:15:53):

In getting the spec of the design, speaking of design the product, not the documentation, to the front end devs, right? Yes. And now you’re basically communicating HTML and CSS. Yeah, pretty much like pixel perfect. What was that like? Because I remember like, you know, back in 2013, we’re using like envision clickable like images, clickable, basically images from Adobe.

Lauren Von Dehsen (00:16:17):

Yeah, I couldn’t even do one better. I think my first design documents were created as if I was making a physical book in InDesign. and then at some point I’d have to print them out and it would be like 150 page stack of paper and what was on which page was kind of a choose your own adventure inevitably an engineer would come and say do you have an answer to this I said oh yeah it’s on page 39 didn’t you see it of course they didn’t see it that’s a

Brian Bell (00:16:44):

freaking textbook

Lauren Von Dehsen (00:16:46):

So it was really like a make do with what you have kind of environment. I think when, you know, Figma came around, we were all trying to wrap our head. If I recall correctly, we were all trying to wrap our head around sketch because that was kind of its predecessor and made this leap to the idea that there could be actually a tool made for this type of design. And we don’t have to just repurpose the tools that are out there for other flavors of design and other media products. in terms of like a Photoshop that would optimize more for imagery and print at the time. So it was really interesting, but I think there was this kind of like, you know, like anything and quite honestly with any other software platform that comes around that we use as technical teams, you have this moment of like, is the migration worth it? Is it worth us all stopping and figuring this out. And so I think there was a lot of mixed yes and no’s on Sketch and trying to figure out what it would really get us, but a lot of people very excited about the potential. And what I remember is Figma just came in on the heels of that so quickly and just had enough of the pieces that were missing, the cloud hosting, the comments, the ability to really collaborate through the tool and not just be in this endless versioning and file upload type of situation. And that was just such a game changer And then I think from there, you know, I think it’s allowed the discipline to mature Because Figma came along at the time where we started to have some ideas around the best way to handle mobile devices And we started to have some ideas on the best web practices And we coalesced on patterns that prior to that we were all trying all different things in each product And so Figma being able to be a shortcut to some of those patterns and ability to not have to spend time on what at that point was the easy stuff and get to really just spend time on the core of the product and the functionality unique to that product. I think it was just a huge accelerant. Now, the challenge on my side is by the time Figma came around, I was more and more and more in a management role. And so I’m really good at commenting. and I’m really good at breaking some people’s files to make a quick edit that I’m trying to communicate. But it is amazing how quickly, you know, you have to really keep up with the tools and learn the tools if you want to stay in it. And something that I’m now looking at, you know, with the new landscape, well, which tools do I want to start learning alongside everybody and not necessarily have a disadvantage there?

Brian Bell (00:19:11):

Yeah, that’s a great segue to the next question, which is sort of, you know, how is the landscape and tools changing with AI? Is there the, you know, the open claw or clawed co-work of design out there like how is that changing how you guys do things because you can if you’re a PM you could just go I don’t know say make me an interface that does this to this and that and this other thing and I’ve done it right yes you’re like a lot of people have had that vibe code experience how’s that changing design as a as a function and are there like new tools that you guys are evaluating that you’re excited about

Lauren Von Dehsen (00:19:43):

Yeah, there’s at a minimum designers are getting the same sort of productivity improvements that everybody’s getting right because there’s plenty of things that we do that’s beyond the pixels or the screen development right that these tools are just accelerating so We’re all in it together. But I’ve been thinking a lot about it lately. I’ve been trying to use it for my own work and I’ve been trying to be more of a, I would say like an IC mindset in approaching these tools. And one of the things that’s become very, very clear to me is that when your work product, hopefully this will make some sense, when your work product was the output of keyboard keys, abstracting that to a different layer of written conversation is a pretty natural progression. It’s the type of abstraction we are used to going through as the tech world. When your output is about moving things spatially or adjusting a color, or trying to create more breathing room and white space or thinking about a sequence of things in more of a flow and a logic or animation for that matter. It doesn’t have as much of the natural abstraction to like trying to describe that in words. And so what I’ve noticed is that the tools are doing a great job of brainstorming. They’re doing a great job of getting really high quality concepts and prototypes that would have taken designers a long time to make and maybe wouldn’t have explored as wide of a set of options in the past because of the time limitations or expediency. Once you get that first thing back, when you try to start getting it closer to what you might have in mind or where you see optimizations, there’s this kind of Fatiguing element the more turns you take and the more specific you’re trying to get the AI to change it almost seems like the harder it is and so I’ve been watching and hopefully I have no preference in this. I have no personal stake in this, but I’ve been watching what Claude is doing now with Claude Design, which is in a very, very early state and has some kind of limited access to it right now. And they’re starting to use a dual interface where you can manipulate some things in line and then you can chat like we’re mostly accustomed to at this point. And I’m really interested in a company called Noon that just came out of stealth mode. It seems like most people are waiting on the waitlist. We’re not sure if anyone’s gotten access just yet, but similar in thinking about how to bridge kind of what a designer needs to where some of the rest of the tools are going. And I’m sure there’s more. I think even I haven’t gotten a chance to go into depth, but I think even Figma made an announcement in the last 24 hours. So I’m hopeful that this is all just a matter of time and it’s just sequencing and what the teams can get to. But it’ll be interesting to see how the tools kind of address a more maybe specific and unique type of working style that a designer brings to the table than some of their cross-functional peers.

Brian Bell (00:22:44):

Yeah, I guess you could say that Figma was kind of the culmination of Web 2.0 tools, right? interactive collaborative user-generated stuff that’s in a web interface and we clearly have not reached the culmination and peak of AI co-pilot driven design yet maybe it’s cloud maybe it’s something else but something will come along where it’s like okay and I’m reminded based on what you said earlier about you know a picture is worth a thousand words you know the old cliche right and a design is kind of worth it’s you know a thousand words too and a video walkthrough is probably worth ten thousand words let’s talk about your SoFi experience. You were there for a long time. Are you still there?

Lauren Von Dehsen (00:23:21):

I just stepped away at the end of March.

Brian Bell (00:23:23):

Wow. Congrats. I mean, you had a very long run. You joined right before COVID, kind of broke everything down. What was that run like? Because SoFi had a really incredible run and you had your hands probably on everything there.

Lauren Von Dehsen (00:23:34):

Yeah, it was a very fast paced and rewarding chapter. I had the insight that it was time to leave Google and I wanted to go to a company at a certain size and phase. And so far was that in the 2020 time period.

Brian Bell (00:23:54):

You mean you didn’t want to go back to another big political, big tech company?

Lauren Von Dehsen (00:23:58):

It’s not where I do best. I have a lot of respect for those that do, but.

Brian Bell (00:24:03):

Yeah, I’m not a politician. That’s my problem.

Lauren Von Dehsen (00:24:05):

Yeah.

Brian Bell (00:24:06):

I say exactly what I mean and I have no filter. And so Microsoft was not a good place for me. People kind of do all the like doublespeak corporate things. And I’m just like, well, I think you’re wrong.

Lauren Von Dehsen (00:24:19):

I can relate to that I can relate to that I don’t have much of a poker face I’ve been told I liked I liked I always liked the teams that felt the level of urgency to make decisions and ship and I found that that came with like a certain stage in the company’s life cycle and yes you could experience that inside of a big

Brian Bell (00:24:37):

company if that organization is very much like that I mean Amazon felt like a startup you know just felt like a series bc startup just moving as fast as you can everybody’s like speaking their mind there’s no pot like there were politics just 10 years ago it didn’t completely different from Microsoft

Lauren Von Dehsen (00:24:54):

Yeah well that I’ve never been inside of Amazon but I could see that based on the cultural tenants so so if I met a lot of the criteria in terms of like scale and phase and it was a totally different problem space which I maybe the history you read out maybe explains I actually like switching after being in a space for a reasonable amount of time and really investing in it I like switching into a new space because I find it brings a lot of creativity

Brian Bell (00:25:21):

yeah it’s like if you’re yeah like you don’t want to sit there and design the next iteration of the thing you’ve already been working on for multiple generations I’m guessing as a designer you’re kind of like all right it’s just another version of Nest you know like great it’s a little more it’s a little more spherical now like you know ready to move on you know

Lauren Von Dehsen (00:25:39):

And the act of learning, I mean, I don’t, every designer is going to be different, but the act of learning a new space and asking all the questions is just a very creativity kind of inducing mode to be in. So the first couple of years of being in a new space, I always found very exciting.

Brian Bell (00:25:56):

Yeah. That’s why I like being a VC. Like I’m constantly talking to founders and doing new things all the time.

Lauren Von Dehsen (00:26:02):

Yes, exactly.

Brian Bell (00:26:03):

So it’s never a dull moment. It’s like, oh, like I didn’t know you could do that with like dermatology, you know? exactly I would have ever gone to YouTube and said show me like what the latest AI is like being used in dermatology you know I’ve never you know that’s right so

Lauren Von Dehsen (00:26:22):

going to SoFi was you know I had I always had interest in my personal finances I felt like they were setting themselves up to be in a spot that could really be interesting across what was a very active fintech environment in 2020. Yes, going just, what, six weeks before we all ended up at home was a very strange and unpredictable circumstance. I was very fortunate to meet people in the office before that started because I think... Yeah,

Brian Bell (00:26:47):

like a month before, yeah.

Lauren Von Dehsen (00:26:49):

Yes, just even having one moment of FaceTime was meaningful. But I think what... That chapter for me, and I’ve done a lot of reflection in the last couple of weeks in terms of stepping away, it was really about how do you create a more grown up and mature research and design organization? And I had come from these environments where people were teaching me what that meant. And I was really lucky to work with groups that were coming from environments that had worked with the best and were doing it at such a high level. And then this opportunity was like okay what have you learned and how do you put it into practice and how do you take a company that wants to grow into that and take them to the steps to do it so a lot of my role was really focused on how to build the discipline and the practice but of course how to execute and keep pace with a growing and changing company and so yeah we we spend a lot of different financial segments from banking and investment and more recently reentering the crypto space but also the bread and butter of the company historically was loans and there was a play to move into insurance so talk about you know all different concepts that some of the other fintechs in the space especially in the earlier years would have only tackled one of those and yet we were trying to tackle all of them in parallel and so a lot to wrap your head around and a lot to kind of figure out as a designer what’s shared and what needs to just be the same because you’re under one brand and one company and one app providing one experience to the customer and then what are all of the variants and specifics that the products just work different and the verticals just work different and you still have to accommodate all of that in its gory detail so it was an interesting kind of ecosystem problem to tackle

Brian Bell (00:28:30):

Yeah, as an investor, I love teams that have, you know, a builder, like a CTO kind of person, you got the CEO, the salesperson, but I love seeing a designer on the founding team. I think it’s a really neat thing when they’re, you know, thinking about design right from the beginning, or like it’s one of the first hires. What do you think founding teams and founders get wrong about design when they’re going from zero to one or one to 10 at their startups?

Lauren Von Dehsen (00:28:58):

When you’re not a designer, you’re coming to it as a founder, a cross-functional peer. It’s easy to have a definition of design that is kind of based in whatever you’ve seen up until that point. And maybe you’ve seen something really amazing and you’re trying to replicate that one-to-one.

Brian Bell (00:29:15):

You mean every startup has a chat interface now, like right down the center of their screen? Yes. Yes But for you know dogs and it’s like this interface it’s right there like see you just like have the dog look at it talk yes yes well there is no there is

Lauren Von Dehsen (00:29:32):

absolutely an art to knowing when to use the pattern that everybody is using and when to diverge absolutely I also mean it from a how does the how does the discipline function inside of the company what decisions are they making what do you task the designer with what permission do you give the designer culturally how do they sit amongst whoever else is there at that early stage it’s an interesting one it’s like maybe an engineer would say that they’ve experienced all different interpretations from their cross-functional peers but from my stance it seems like the uh the cross-functional definitions of design somehow span a wider breadth. And so...

Brian Bell (00:30:11):

There’s gotta be like a design meme out there where, you know, like you got the product managers like, But do our customers want it? And then, you know, the engineer is like, but can it scale to a million users? And then the designer is like, but is it like easy and fun to use? Right. And it’s like almost like the triangle, you know, cost scope and time triangle. Yes.

Lauren Von Dehsen (00:30:30):

Yes. Yes. And I found, you know, like I mentioned before that Nest was one of the chapters that I often go back to as just amazing learning experience. I found that in that environment, I never had to explain what design did or what why I should be in that conversation or what I had to contribute.

Brian Bell (00:30:46):

That was the differentiator of the company, right? Just better design everyday products, you know?

Lauren Von Dehsen (00:30:51):

And the way that that was achieved was that every function cared about design and every function understood that they were making decisions that were contributing to the experience and what other companies might say were contributing to the design. And the best way that I could explain that set of cross-functional partners was that anyone could argue any element of the product development. It didn’t have to be in their lane. I had plenty of people tell me what feature they wanted to see next or why the feature should work differently from a design perspective. But once you heard everybody’s point of view and you took in all the feedback and contribution, the team really trusted whoever was sitting in that seat to make a final call as a subject matter expert. So you had this constant working in the gray space, everyone working in the gray space, and then this trust that the right person with The decision making responsibility would ultimately make the right decision. And I think that was like the most maybe inspired and interesting definition of design where everybody felt responsible for it, but yet allowed the designer to make a call. Whereas, you know, I think where SoFi was in its maturation process, I was coming into a team where we want something different out of the design function and we want it to be more strategic and we want the types of results that would come from a strategic minded design team. but we’re in a situation where the way information is getting to them and the types of tasks we’re tasking them with are far more boxed in and without context and less ownership and some elements of that is oh is the team at the right seniority is the team at the right skill set mix do you have the the right people doing the right types of things like engineering there’s many many different sets of skills and jobs within a design team even if they have a design title they might not be coming to the table with contributions and a design team knows how to use them. But cross-functionally, do you know if you got the right type of person? So when you’re early stage as a founder, I think the question becomes, what are you looking for this to do? How are you thinking about it in the mix of where you’re starting? Is this something you want to make as a tenant of your company? And if so, are you appropriately finding someone with seniority and experience the way you would for other disciplines? Is this more of a means to an end to get to an MVP? It doesn’t mean that any of those are right or wrong. Does your decision match what you’re trying to achieve and what you think is important to that business and that product?

Brian Bell (00:33:25):

Let’s wrap up with some rapid fire to wrap up questions.

Lauren Von Dehsen (00:33:29):

Sounds good.

Brian Bell (00:33:30):

I hesitate to call them rapid fire they’re kind of like kind of just loose ends a loose end section so you wrote that documentation will never be seen by the user only the product matters apply that to design leadership what’s the equivalent of over documenting and how design leaders run their organizations I think the

Lauren Von Dehsen (00:33:44):

equivalent is probably the way the design leaders set up the rituals right in the same way rituals are there to facilitate work and connection and feedback and yeah exactly make some amount of predictability in what’s otherwise a very organic and somewhat chaotic process yeah And so one of the things we did a lot with at SoFi sometimes successfully sometimes was a learning learning misstep that we needed to improve was at each kind of change of scale we had to look at are the rituals still working for us or does it have to adjust is the test the team gotten too big for this format to work and so

Brian Bell (00:34:22):

being really having like design committee meetings with 30 people and I’m like oh yeah this is probably too many people giving feedback on this experience

Lauren Von Dehsen (00:34:30):

Exactly there’s a tipping point of how many people can be in an effective critique where you’re really helping each other as peers then there’s a purpose to do cross-functional reviews and do you have the right people in those meetings and has that changed from when they originally got set up because the dynamics of the company have changed with scale and evolution and so I think as a leader you have to be willing to say this is the process I like to use and I’m going to bring my expertise to the table but I also need to be watching what’s working for the team and at some point the thing I love using may no longer be the right call and does somebody else have a good idea of what we do next or have I thought about it deeply enough that I have an idea of maybe what we’ll try in the next phase but it’s that facilitation it’s like the facilitation can’t be the thing The facilitation has to be the means to the outcome that you’re really trying to focus everyone on.

Brian Bell (00:35:19):

So you worked on Matter. Maybe you could explain what that is. And it’s a rare thing because it kind of outlived your tenure at those companies and now it kind of lives on in multiple companies.

Lauren Von Dehsen (00:35:30):

The thing I worked on at the time was called Thread and Weave, and it was a protocol, networking protocol that Nest created as kind of a necessary component to how the devices were meant to work with each other. And it created a bit of a closed loop system where anything in the house could communicate over a different type of network, what we would have called the mesh network. and it was neither Wi-Fi nor Bluetooth, which like blew my mind when I got there. Somebody tried to explain that to me and I said, what do you mean there’s a third thing?

Brian Bell (00:36:01):

It’s like a whole nother radio frequency. Don’t worry about it.

Lauren Von Dehsen (00:36:04):

Exactly. And I really thought the person didn’t know what they were talking about for the first 20 minutes of the conversation and go, wait, I don’t know what I’m talking about. So it was really a fascinating project to work on. And what eventually happened is we all knew that with these home devices, locking somebody into one environment just didn’t seem like it was going to be practical over a long period of time. But when I was working on it, we didn’t know how we were going to get there. And the team that lived on with it eventually turned it into what we now call matter and I’m sure it’s evolved quite a bit but it was fascinating because it was true technology innovation and when you’re thinking about radio frequencies and mesh networks and yet part of the reason I was hired was because I had worked on the onboarding and pairing elements of the fuel band. And so I knew that it was a totally different problem when I got in, but from the outside, it looked like kind of the same design problem. And that was how my involvement started there.

Brian Bell (00:37:04):

So you’ve been around a lot of different teams and we talked a little bit about founders and setting up kind of design cultures, but what is a single most common design mistake early stage companies are making and a thing that you’d fix in the first 30 days if you went into like an early stage startup?

Lauren Von Dehsen (00:37:18):

My honest answer is I would assess and I would try to figure out what I thought maybe the bottleneck was. But one of the common things that I would imagine might be worth fixing early on would be how is the start of a project developing? It’s really easy to bring in each function at the point at which it’s obvious that function needs to contribute. But I think the best products and the best teams actually think about over communication and inclusion really, really early in a project before it’s obvious that all of those people need to be in the meetings and tracking along. And I think as a designer, especially if you have a really talented design team and really capable individuals on the team, the more context you can get them in the beginning of a project, the more that they can really be utilized in the fullest sense. And so that’s something that I tend to look for.

Brian Bell (00:38:09):

Design communities tend to celebrate process sprints, design thinking, double diamonds, qualitative panels. So you published a piece that says process is a trap and the product is the only thing that matters. That was a long time ago. Has your view evolved or do you kind of stick to that point?

Lauren Von Dehsen (00:38:24):

I think it depends. I still think some of the language we use to talk to ourselves falls flat. with our partners. I think we can be at times, and I apologize to the design community, but we can be like a little bit too esoteric about things and think that that’s going to be what makes it through when really our partners just want to know when we’re going to have the answer and if it’s going to be by the deadline they’re expecting.

Brian Bell (00:38:50):

Yeah. When will you have the design?

Lauren Von Dehsen (00:38:52):

That’s right.

Brian Bell (00:38:53):

Yeah. We need it. We start our sprint Monday and it is Friday afternoon.

Lauren Von Dehsen (00:38:56):

That’s right.

Brian Bell (00:38:57):

Yeah. And so those can you have it done by Monday morning? Yeah, like I remember all these conversations.

Lauren Von Dehsen (00:39:03):

Yes and there’s a whole other piece right where you have to set realistic expectations and negotiate about those types of things but you know I think those tools serve a purpose when it’s the ritual or the approach that gets to the conclusion or the next step that the team is trying to unlock but it’s not because they have a fancy name and a fancy concept behind it and that everybody participating in that ritual needs to fully understand it it’s because it might be the right tool for the job And so, you know, we in more recent years, we pushed really hard on design sprints. And one of the conversations that would happen internally in the design leadership team would be like, some of these things we’re calling a sprint isn’t really a sprint or it’s some other form of a sprint that we would call something totally different.

Brian Bell (00:39:48):

It could just be a block of work, you know?

Lauren Von Dehsen (00:39:51):

Right.

Brian Bell (00:39:52):

And so work.

Lauren Von Dehsen (00:39:53):

so I mean they probably they’re probably sick of me saying this but I said look right now all of our partners understand what a sprint is and they are bought into it and they’re seeing the results of it. And we want to keep doing these. Now is not the time to get technical about the name and which variant we’re doing when and why this is fit.

Brian Bell (00:40:14):

Do I have to work faster and sacrifice quality in a sprint? Designers are like really thinking it through, right? Yes. What is the essence of a sprint? Let’s like talk about it.

Lauren Von Dehsen (00:40:25):

Yes we like to we like to noodle on those details and that is part of what we do but when misplaced it can be counterproductive and so I’ve just never subscribed so much to the precision of the names and the kind of on paper process and it’s more about which thing is going to facilitate what we need in the moment and try to be a little bit more organic and flexible with the process What’s something you’ve

Brian Bell (00:40:50):

changed your mind on that you like used to believe that you no longer believe or vice versa

Lauren Von Dehsen (00:40:55):

That’s a good one. I think when it comes to design, I think that I’m a little bit more understanding and flexible to meet the business where the business is. I think when you start out in design, you want everything to be at the most resolved, beautiful, elegant,

Brian Bell (00:41:16):

thought out. The further you go from the code through design, through product, through the rest of the organization, the messier things get. yes you know like up here in the business world it’s messy you know lots of politics and different product lines and different you know just all kinds of different constituents and different needs and functions absolutely a little bit more pure design is like you know the the physics of products right what I mean by that is like mathematicians kind of shit on physics people and vice versa like you guys are like this is how things work and this is like and then and then along come the coders and I know this is like math it’s pure it’s code right and it’s never heard it put that way but I follow you I follow you I just this is what I do I just come up with stuff on podcasts and I get paid for it I guess now

Lauren Von Dehsen (00:42:08):

Yeah but yeah I mean you start out in design wanting to make the world beautiful right and do the best you possibly can and it’s not to say that that part of my opinions changed but what’s appropriate for the problem being solved and where is the perfection, maybe a necessary trade-off to efficiency or speed to market or some other thing that’s going to be highly critical and important to the product and the business success. And I think I had a really early awakening to that that kind of started the path of changing my mind when I started managing where there were times I had to make a call against my team’s recommendation. And it wasn’t because I didn’t understand the team’s recommendation or they weren’t. Well reasoned in the recommendation but I was responsible for working cross-functionally to understand how that stacked up against other constraints and requirements and sometimes a different requirement had to win out and so it is interesting to have to kind of learn that lesson and then once you do find a way to apply it where you’re not losing your standards or expectations but applying it at the right times.

Brian Bell (00:43:19):

Yeah. Well, that was a really fun conversation. I learned a lot and really enjoyed it. Thanks for coming on.

Lauren Von Dehsen (00:43:25):

Absolutely. Thank you so much.

That number traveled fast, and it earned the trip. It is the largest single equity round anyone has ever pinned to an artificial intelligence lab. If you wanted proof that the money still believes, there it was in nine digits.

I spent about twenty minutes on it, then moved on to the item that mattered more. It was smaller, it was quieter, and it ran underneath the headline like a current under a calm surface. Two of the biggest investment firms in the world, Apollo and Blackstone, were lining up roughly thirty-six billion dollars in loans. Not to buy stock in Anthropic. To buy computer chips for it.

Here is the shape of the deal, in plain terms. Apollo and Blackstone run what is called private credit, which means they lend money directly to companies instead of letting a bank stand in the middle. The thirty-six billion would buy a large supply of Google’s specialized AI chips, the processors Google designs in-house to run models like Claude. Anthropic would then lease that hardware, the way an airline leases planes rather than buying the whole fleet outright. And Broadcom, the company that helps design those same chips, agreed to stand behind about thirty-one billion of the loans. If the chips end up worth less than the lenders are betting, Broadcom absorbs part of the loss.

Read those two announcements side by side and the week tells a different story than the headline did.

Why the quiet number matters more

There is a real difference between raising money by selling a piece of your company and raising it by borrowing. When you sell equity, a slice of ownership, you hand over part of the upside and you owe nothing back. If the bet goes wrong, your investors lose with you. Patient money. The most patient money in the world, in fact, has been what built the AI labs so far.

Borrowed money is a different animal. You owe it back whether the bet works or not. You owe it on a schedule. And what you owe moves with interest rates, so the cost of the whole thing depends on conditions you do not control. For most of the AI boom, the enormous bills for chips and data centers were paid with equity, the patient kind. This week marked the point where the bills got big enough that even Anthropic’s near-trillion-dollar valuation could not cover them with stock alone, so the buildout started leaning on debt.

That changes the risk in a way the sticker price hides. A company financed entirely by equity can survive a bad year by simply having a bad year. A company carrying tens of billions in lease obligations has payments to make in the bad year too. The AI buildout just acquired a clock.

The Broadcom piece is the part I keep turning over. A chip designer promising that chips will hold their value is a company underwriting demand for its own product. That works beautifully while demand is roaring, and demand is roaring right now. It becomes a problem the moment it stops, because the same firm is exposed twice, once as the seller and once as the guarantor. None of this is reckless on its face. It is the sort of arrangement that looks obviously fine for years and then looks obviously foolish in about a week. Experienced investors have watched that movie in other industries, and the ending depends entirely on whether the demand was real or borrowed.

The timing is the tell

The same morning the debt deal surfaced, the government published its main inflation report, and it did not cooperate.

The Federal Reserve watches one inflation gauge more closely than any other. It goes by an ugly acronym, PCE, and the version that strips out volatile food and energy prices came in at 3.3 percent for the year through April. The Fed wants that figure near 2 percent. It has now been stuck well above target long enough that the word “transitory” has quietly retired. At the same time, the government revised its estimate of early-year economic growth down to 1.6 percent, slower than first thought. And the share of income people save fell to 2.6 percent, which means households are dipping into savings to keep spending at the level they are used to.

Put the two stories together and the tension is hard to miss. The AI industry started borrowing heavily to build, right as borrowing stayed expensive and the broader economy began to soften. For most of the past two years, a lot of venture math quietly assumed that cheaper money was coming, that rates would fall and make every aggressive plan look smart in hindsight. After this week, any founder still leaning on that assumption is leaning on a chair that got removed from the room. The rate-cut rescue is not arriving in 2026.

The rich got richer, on schedule

Step back from the debt deal and the rest of the week rhymed with it. Money kept piling into a small number of proven names while everyone else waited in line.

Anthropic’s raise looks slightly less colossal once you read the fine print. Roughly fifteen billion of the sixty-five was money its big cloud partners had already promised, so the genuinely new equity is smaller than the headline implies. Still enormous. Just not quite the figure that traveled.

Cognition, the company behind an AI software engineer called Devin, raised more than a billion dollars at a valuation of twenty-six billion, more than double where it stood eight months earlier. To see why that number raised eyebrows even in a giddy market, you need one concept. Investors often price a company as a multiple of its revenue, so a company earning a hundred dollars a year and valued at a thousand is trading at ten times revenue. Cognition was valued at roughly fifty-three times its revenue. The company reported it now runs at a pace of about 492 million dollars a year, up from 37 million a year earlier, which is a genuinely wild rate of growth. Fifty-three times is still a price that assumes the growth continues without a stumble. The detail I found most telling sat off to the side. Cognition says its own AI now writes more than 90 percent of the company’s code. The product builds the company that sells the product.

At the other end of the market, Meta did something that squeezes from below. It started selling AI subscriptions for $7.99 a month, with a premium tier at $19.99, testing first in Singapore, Guatemala, and Bolivia. The incumbents charge around twenty dollars for the same general idea. Meta is undercutting them by more than half, and it is doing it on top of an app empire with something close to a billion people already using its AI. So the week pressed startups from two directions at once. Giant rounds pulled talent and capital toward the top, and giant distribution pushed the price of basic AI toward the floor. The comfortable middle, a startup charging twenty dollars a month for a clever general-purpose assistant, got thinner from both ends.

A different track: the machines kept getting smarter

Everything above is the money story. There is a separate story running alongside it, and it is worth keeping the two apart, because they move for different reasons and reward different things.

The capability story had two moments this week.

The first was a new version of Claude, Opus 4.8, which Anthropic released on the same crowded May 28. The interesting claim was not that it scored well on coding tests, though the company says it did. The interesting claim was that this version is also its best-behaved, meaning less likely to do something its makers did not intend. For years the working assumption inside AI was that making a model safer made it a little dumber, that you traded capability for control. Anthropic is now claiming the opposite happened, that the strongest model is also the most reliable one. If that holds up under outside testing, and the right move is to treat any company’s claims about its own product as unproven until others reproduce them, it flips a long-standing tradeoff. Reliability stops being a tax on intelligence and starts being a feature you can sell, especially to banks, law firms, and anyone whose real fear was not that the machine was slow but that it would confidently make something up.

The second moment was stranger and, to me, more important. A team of four mathematicians published a paper disproving a long-standing conjecture in number theory, a problem about whether certain sets of numbers can stay small under both addition and multiplication at once. What makes it remarkable is how they did it. They borrowed a technique that an AI system had produced just one week earlier while cracking a different unrelated problem. The humans read what the machine did, recognized the method could travel, and carried it into their own field in days. An idea discovered by a machine seeded real human research almost immediately.

That is the development I would underline twice. We have spent a couple of years asking whether AI can do impressive things on its own. This was something else, a machine and a room full of humans trading techniques back and forth fast enough that the boundary between who discovered what started to blur. If that loop keeps tightening, the old comfort that frontier knowledge diffuses slowly, that a clever insight gives you years of advantage before the world catches up, gets shorter every cycle.

How this changes the way I read a valuation

The lasting lesson of the week is not any single deal. It is a habit of reading.

When a company is financed entirely with equity, the headline valuation tells you most of what you need to know about how the market prices it. When a company is financed partly with debt, the headline tells you less, sometimes much less, because the real obligations are sitting in a financing structure that the equity price ignores. Anthropic’s near-trillion-dollar valuation does not include the tens of billions in chip leases stacked behind it. Anyone trying to judge what the company is truly (net) worth has to add that weight back in, the same way you would not value a house by its purchase price while ignoring the mortgage.

That habit will matter well beyond this one company, because the debt deal is almost certainly a template, not an exception. The sums required to build frontier AI have outgrown what equity markets will hand over, and private credit is stepping into the gap. From here forward, reading an AI valuation means asking what is borrowed, who guaranteed it, and what happens to the payment schedule if demand cools or rates stay high.

If you want the short version, here are the questions I am carrying into next week:

• When a company brags about a valuation, how much of the financing behind it is debt, and who is on the hook if the bet sours?

• If basic AI keeps getting cheaper at the consumer level, what does a startup own that a giant with a billion users cannot simply copy at a lower price?

• Does the business survive if the cost of computing stays flat or rises, rather than falling the way everyone has been quietly assuming?

The boom is not slowing down. What changed this week is the kind of fuel it is running on. For two years the AI buildout was financed by people who could afford to be wrong. It is starting to be financed by people who cannot, on a schedule that does not care whether the future arrives on time. That is a different machine than the one we have been watching, and it is worth knowing which one you are looking at before you decide what it is worth.

The models are Anthropic’s, the family called Claude, and lately the release schedule has been doing something that gives me a small case of vertigo. Version 4.5 arrived in late November. Then 4.6 in early February. Then 4.7 in the middle of April. Then 4.8 this week. The spaces between them keep narrowing, from roughly seventy days down to forty-two, and a larger model sitting above the whole line, one that has been running quietly inside a few dozen companies for a couple of months, is said to be weeks away from reaching the rest of us. Every release is better at the work I hand it. Every release makes the version I just figured out a little obsolete.

There is a name for the feeling, and it turned fifty-six this year.

In 1970 a writer named Alvin Toffler published a book called Future Shock. He defined the title phrase plainly, as the distress people feel when they are subjected to too much change in too short a time. The danger he was pointing at was never a particular invention. It was the rate. Toffler argued that the speed of change had crossed a threshold where human beings could no longer process it, and that the result would be a population that felt permanently disoriented, off balance, unable to find footing because the ground kept moving under them.

He had a device for making the speed legible. Picture the last fifty thousand years of human existence divided into lifetimes of about sixty-two years each. You get roughly eight hundred of them. Only in the last six did large numbers of people ever see a printed word. Almost everything in the room you are sitting in was invented in the most recent one, the eight hundredth. Toffler wrote that in 1970. Someone made the point recently that we are now living at the start of the eight hundred and first.

Toffler is also the man who took the phrase information overload and put it into ordinary speech. He predicted that people drowning in input would start making worse decisions, would retreat into rigid routines, would cling harder to old certainties, and would grow suspicious of the experts who kept assuring them the future was under control. Read that sentence again and notice how little of it needs updating.

Toffler got the diagnosis right. He got the outcome wrong, and the way he got it wrong is the part worth sitting with. He thought the overload would break us. He used the phrase adaptational breakdown and meant it close to literally, a society of people made sick by the pace, unable to cope. That did not happen, or it has not happened yet in the way he imagined. The rate of change kept climbing and most people are, by the ordinary measures, fine. They adapted. They always do.

What happened instead was quieter and in some ways harder to live with. The pace did not break everyone evenly. It sorted them. Some people and some companies turned out to be very good at absorbing change quickly, throwing out what they learned last month and relearning, and those people pulled ahead at a speed that would have looked like cheating a decade ago. Everyone else kept running and stayed in roughly the same place. The shock did not produce collapse. It produced a widening gap between the fast and the rest, and the gap is the thing that destabilizes a life, because you can feel yourself falling behind even while you are working harder than you ever have.

We have started calling the present moment the singularity, a word that used to belong to physics and science fiction and now turns up on earnings calls. The original idea was a point where machine intelligence starts improving itself faster than we can follow, after which prediction stops working. Sam Altman, who runs OpenAI, wrote last summer that we are already past the event horizon, that the takeoff has started, and that so far it has felt less strange than people feared. Ray Kurzweil, who has been making this forecast since the late 1990s (really the 80s) and has been right more often than he has any business being, still puts human-level machine intelligence around 2029 and the full merger of people and machines around 2045.

Then there are the people who think this is mostly a story we are telling ourselves. Yann LeCun, one of the researchers who built the foundations of modern AI, calls the timeline overhyped and argues that the systems everyone is marveling at do not reason in any deep way. Gary Marcus, a cognitive scientist who has spent years as the loyal opposition, likes to point out that these models can recite the rules of chess flawlessly and then, a few moves later, slide a piece across the board in a way the rules forbid, because they never built a real model of the game. They match patterns. Sometimes the pattern breaks.

Both camps are serious, and both are partly right, which is the only honest place to stand. The singularity as a felt experience is here. Anyone trying to stay current with what these tools can do has been living inside Toffler’s too much change in too short a time for a year or more. The singularity as a technical event, machines bootstrapping themselves to something past us, remains a forecast. Smart people are betting careers and billions on it, smart people are betting against it, and the predictions that were supposed to land by now have a habit of sliding a year into the future every time the calendar catches up to them.

The next step in the story, the one I think is still genuinely far off, is wiring the machine straight into the person. Brain-computer interfaces. The progress is real and worth respecting. Neuralink has put its implant into a couple dozen people who could not move, and they are steering cursors, typing, playing chess, in one case operating a robotic arm well enough to feed themselves. A company called Synchron threads its device in through a blood vessel in the neck so it can skip open-brain surgery entirely, trading resolution for a path that might actually reach a lot of patients. For someone living with paralysis, this is close to a miracle and it is arriving now.

The dream people reach for when they hear about it, plugging a healthy brain into the network and thinking faster, sits on a different timeline. Nobody has done it. The technologists who talk about it openly put it in the 2030s and 2040s, and they tie it to hardware that does not exist yet. So when I say the next move is augmenting our own minds directly, I mean it the way you mean a city on Mars. Plausible, underway in its earliest form, and further off than the excitement around it suggests.

I want to bring this down to the work I do, which is putting small amounts of money into very young companies and trying to guess which ones will matter. The acceleration is not abstract there. It is the whole texture of the job now.

A few years ago a software company with ten or twenty people and ten or twenty million dollars in annual revenue was a serious accomplishment that took years to build. Garry Tan, who runs the startup program Y Combinator, said last year that he is now watching teams that size reach those numbers in well under two years, and that his most recent groups of companies have been growing about ten percent a week across the entire batch, something he had never seen in his career. A coding tool called Cursor reached half a billion dollars in annual revenue with fewer than fifty employees. One of the better-known image companies runs on a team you could seat around a dinner table. Dario Amodei, who runs Anthropic, was asked when we would see the first company worth a billion dollars run by a single person, and he answered 2026 (one could argue this already happened with OpenClaw).

So here is the comfortable version of the thesis, the one that feels good if you are on the right side of it. The founders and the investors who win from here are the ones who metabolize change the fastest. They adopt the new tool the week it ships, rebuild around it, and move before the advantage evaporates. Speed of adaptation decides everything.

I believe a version of that. I also think it becomes a trap the moment you stop there, because the same speed that mints the winners is quietly dissolving the instruments we used to tell winners apart. For my whole VC career, reading a young company meant leaning on a few reliable signals. How many people did it take to build this. How much money are they burning. How fast is revenue growing and how steady is it. Those numbers meant something because they were expensive to fake and slow to move. Now a tiny team with the right tools can produce numbers that used to require a real organization, which means a chart that would have made me reach for a checkbook in 2018 tells me almost nothing on its own today. Some of that revenue is durable. Some of it is people trying the shiny new thing for a month and leaving. On a spreadsheet the two look identical.

And the failure mode of pure speed is moving fast in the wrong direction while certain you are right. I have watched founders use these tools to manufacture a year of apparent traction in a few months, and in more than one case the corner they cut to get there became the thing that sank them. A regulator’s letter. A customer who turned out not to be real. A number that did not survive contact with diligence. Adapting quickly to the new pace is necessary. It is not enough. The thing it cannot replace is judgment, which moves at its own stubborn speed and refuses to compress.

Which brings me to a scene from this week that I keep turning over. I have spent years building a model that reads thousands of past investments and tries to predict which new companies will succeed. A few days ago I pointed it at two hundred startups from a single Y Combinator batch and let it score all of them in one pass, work that would once have eaten a whole venture team’s week. Then I sat down and started going through the two hundred by hand, one at a time, checking what the machine concluded, hunting for the place where its tidy pattern slid a piece across the board in a way the rules forbid.

That is the whole thing, right there. The machine gives me reach I never had. I give it the one thing it does not have, which is the willingness to be slow on purpose about the decisions that matter. At the end of his alarming book, Toffler landed somewhere I did not expect. He told readers to build what he called stability zones, places and habits that hold still so the rest of life can move fast around them. I used to read that as nostalgia. I think now it was the most practical advice in the book. You survive too much change in too short a time by choosing, deliberately, the few places where you hold still.

We are at the start of the eight hundred and first lifetime. It is moving faster than the eight hundred before it put together. The people who do well will not be the ones who feel calm about that, because nobody sane feels calm about it. They will be the ones who learn to move fast on the tools and slow on the judgment, and who can tell, in the moment, which is which.

But for many drivers, the future still gets stuck at a broken public charger.

That tension is exactly what Brad Crist, co-founder and CEO of ChargeMate, is trying to solve. After spending years in climate tech and EV infrastructure, including work at companies like Accenture, Faraday Future, Volta, and Spring Free EV, Brad saw the same problem again and again: the industry was great at deploying chargers, but not nearly as good at making sure they actually worked when drivers needed them.

The result is a massive trust problem. EV adoption is not just about getting more cars on the road or installing more charging stations. It is about whether a driver can pull up, plug in, and confidently get back on the road.

Right now, that experience still breaks too often.

Subscribe now

The Hidden Bottleneck in EV Adoption

Brad’s founding story started with a very human moment: a road trip in his Rivian.

He had been an early EV adopter and was excited to show friends what a luxury electric vehicle could do. Instead, the trip exposed the frustrating reality of public charging. New apps. Slow chargers. Damaged equipment. Confusing payment flows. Stations that appeared to be online but did not actually work properly.

That experience sharpened the problem ChargeMate is now focused on: the gap between charger “uptime” and real driver success.

A charger may show up as available in a system. It may technically be online. But from the driver’s perspective, the session can still fail because of payment issues, app problems, charger-vehicle handshake errors, slow charging, connectivity failures, or confusing user flows.

Brad noted that more than 20% of charging attempts fail, a brutal number for an industry trying to convince mainstream consumers that EVs are ready for everyone.

That is not a minor inconvenience. It is a category-level adoption blocker.

ChargeMate’s Bet: AI as the Operating Layer for EV Infrastructure

ChargeMate is not just building another customer support chatbot.

The company is building an AI-powered operating layer for EV charging networks. Its platform uses chat and voice agents to help drivers in the moment, while also integrating into the backend systems that manage chargers.

That matters because fixing the experience requires more than answering basic questions.

ChargeMate can check whether a charger is online, identify faults, inspect transaction status, and in some cases take remote actions like rebooting a unit or releasing a stuck plug. The goal is not just to respond faster. It is to actually resolve the issue.

That is the core wedge: using AI to turn fragmented, unreliable charging support into something closer to real-time infrastructure management.

Why Generic Support Tools Are Not Enough

A natural question comes up: why can’t Zendesk, Intercom, or an incumbent charging network just build this?

Brad’s answer is that EV charging is not normal customer support.

This is a messy intersection of software, firmware, hardware, payments, vehicles, field service, and physical infrastructure. One ChargeMate customer has dozens of hardware products to manage, and one hardware SKU alone has hundreds of unique error codes.

That complexity is exactly where vertical AI has an advantage.

A generic AI support system can answer questions. ChargeMate is designed to understand the specific failure modes of EV charging: the charger, the vehicle, the network, the transaction, the driver behavior, and the operational workflow behind it.

The more hardware types, vehicles, networks, and failure modes ChargeMate sees, the better its resolution engine can become. That creates a potential data advantage that broad horizontal tools will struggle to match.

The Business Case: Better Support, Better Margins

EV charging operators are under pressure to prove that their networks can become profitable businesses.

That is hard when support and operations costs are high, charging sessions fail, and customers abandon the experience. Brad described customer support and operations as a major cost burden for operators, especially when calls can cost $10 to $15 each.

ChargeMate’s early results are meaningful. Brad said the AI can resolve roughly 50% to 70% of calls without involving a human. He also pointed to one client seeing a roughly 5.5% lift in charging success rate, which can translate into millions of dollars of margin at larger network scale.

That is the real investor-relevant insight: ChargeMate is not selling “AI support.” It is selling recovered revenue, lower support cost, better asset utilization, and improved driver retention.

In infrastructure markets, reliability is margin.

The Pivot That Made ChargeMate Work

ChargeMate did not start exactly where it is today.

The original idea was closer to a consumer product: helping EV drivers find available, working chargers near desirable stops like coffee shops or clean bathrooms. The problem was real, but the go-to-market path was ugly.

Competing with Google Maps or trying to become a consumer endpoint would have required massive distribution. Accessing vehicle data would have been difficult. Monetization would have been uncertain.

The sharper wedge was on the operator side.

Charging networks were already paying for failed sessions, expensive support calls, unhappy drivers, and fragmented operations. ChargeMate could solve a painful B2B problem with direct ROI.

That pivot matters because it reflects one of the biggest lessons in startup building: the best product idea is not always the best business. ChargeMate became more interesting when it moved away from broad consumer convenience and toward a painful operational problem with budget attached.

Voice May Be the Real Unlock

One of Brad’s more interesting reflections was that ChargeMate may have started too heavily with chat.

When a driver is stranded at a charger, frustrated, and trying to get moving, the natural behavior is not always to open a chat window. It is to call.

That is why ChargeMate is now investing in voice AI as a major interaction point. Voice fits the urgency of the moment. It also lets ChargeMate take over the first line of support while escalating to humans when necessary.

This is where the company’s hybrid model becomes important. ChargeMate combines AI with human call center partners, creating an AI-enabled service layer rather than a purely software-only experience.

That is likely the right architecture for messy infrastructure markets. Full automation is attractive, but trust is built by solving the problem. Sometimes that means AI. Sometimes that means a human. The winning system routes intelligently between both.

Where the Market Goes Next

Brad sees EV charging as the beachhead, not the full opportunity.

The larger idea is that AI can become the operating layer for energy infrastructure. Chargers are just one class of distributed physical assets that need monitoring, support, diagnostics, and coordination.

Customers are already asking whether ChargeMate’s AI can help manage other systems, including batteries, building management systems, and broader site-level energy assets.

That points to a bigger future: self-healing infrastructure that can detect problems, communicate with humans, coordinate workflows, and resolve issues before they become expensive failures.

For EV charging, that future cannot arrive fast enough.

The Takeaway

The EV industry has spent years racing to deploy more chargers. That was necessary, but it was not sufficient.

The next phase is reliability.

Drivers do not care whether a charger is technically online. They care whether it works when they pull up. Operators do not just need more stations. They need better uptime, better support, better diagnostics, and better economics.

ChargeMate is betting that AI will become the connective tissue between drivers, chargers, operators, vehicles, and field teams.

That is a sharper thesis than “AI for customer support.” It is AI for physical infrastructure reliability.

And if EVs are going to move from early adopters to the mainstream, that may be one of the most important layers still missing.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:
00:01 – Brad Crist, ChargeMate, and the EV Reliability Gap
00:34 – From Utilities to EV Charging Startups
01:16 – The Rivian Road Trip Problem
02:39 – Tesla, Non-Tesla EVs, and Market Fragmentation
04:07 – Why Public Charging Fails
05:55 – ChargeMate’s AI Support Layer
06:18 – Why Incumbents Struggle with Reliability
08:09 – AI-Enabled Support and Call Deflection
09:33 – The First Design Partner Breakthrough
11:16 – The Pivot Away from Consumer Route Planning
13:39 – Why Zendesk and Intercom Are Not Enough
15:38 – Complexity, Protocols, and Charging Standards
16:42 – Autonomous Vehicles and Future Infrastructure
18:11 – Natural Language Interfaces for Energy Assets
19:25 – Early Decisions That Nearly Killed the Company
20:55 – Contrarian Beliefs About EV Infrastructure
21:37 – Enterprise Sales Challenges
22:38 – Starting with Voice First

Transcript

Brian Bell (00:00:57): Hey everyone, welcome back to the Ignite podcast. Today we’re thrilled to have Brad Crist on the mic. He is the co-founder and CEO of Chargemate, an ai platform tackling one of the biggest bottlenecks in EV adoption: unreliable charging infrastructure and broken driver experiences. Prior to Chargemate, Brad spent over a decade in climate tech and helped scale EV charging networks at Volta from hundreds to thousands of stations globally. Thanks for coming on, Brad. Good to see you again would love to get your origin story. What’s your background for the audience?

Brad Crist (00:01:26): Absolutely. So as you introduced us, we spent a lot of time working in the energy industry. I started my career at Accenture with utility clients before moving into EVs and EV charging at startups like Faraday Future, Volta and Springfield. for EV and I was a patient early adopter and early reservation holder of a Rivian which I was excited then when it finally came out to take my partnered friends on road trips thinking I would impress them with this luxury electric car and it seems like every time we’ve I visited a public charger there was some hang up I was downloading a new mobile app to start a charge on a foreign network we were troubleshooting a slow charger reporting damage or wound up in the middle of nowhere and I had been charging electric cars for quite a while but maybe too patient to some of the challenges that exist for the mainstream driver and that road trip really exposed the challenges for us so We focused on solving for the discrepancy between uptime, charging assets look like they’re online, and then the actual experience drivers have is very different and more than 20% of attempts fail. So that’s what we’re after fixing.

Brian Bell (00:02:33): That’s crazy. I mean, one out of five attempts fail. And I love this visceral story. of your brand new shiny Rivian, you know, and like you’re driving around and it’s the R1S, you know, the SUV.

Brad Crist (00:02:45): That’s right.

Brian Bell (00:02:46): So it’s a luxury SUV, right? It’s like a $100,000 car and you pull up to some charging station, you know, you’re down to 20%, 30%. I got to charge and one out of five times that’s not going to work and everybody I think anybody listening who has a non-Tesla anything Neve doesn’t start with the letter T has experienced this problem and even me like I had my first EV was an Audi at the little A3 and Experienced this problem so much and I was so happy to finally get a Tesla. You don’t have that problem but what percentage of EVs out there are non-Teslas now?

Brad Crist (00:03:18): Great question. Tesla has about 40-50% of the market but that’s shrinking. Yeah. And we’ve actually seen the Tesla experience is starting to get worse as they open to non-Tesla vehicles. There’s also something interesting happening where they’re franchising Tesla for third parties. So we’re really curious what that means from a quality perspective. You’ve got a lot of other smaller hosts using Tesla equipment with a lot of other non-Tesla vehicles.

Brian Bell (00:03:44): I’m guessing if you’re a Tesla owner and you pull up and I see a bunch of Rivians parked in all the spots, I’d be pretty mad. Yeah, there’s a little bit of...

Brad Crist (00:03:53): This is ours, like loyalty to the brand. And so there’s some locations that are open to non, like multiple vehicles, others that are Tesla only, which adds a little bit of wrinkle on like, well, where can I visit?

Brian Bell (00:04:05): Yeah. And so what is it about how ChargeMate works? So we get the problem, right? One out of five charging experiences fail initially so that’s a hair on fire problem right for both the ED owner but also the various charging infrastructure providers and then you have the story of like having to download an app every time you stop to charge like a new app a new like my credit card and like oh like Chase now wants to make sure it’s actually my credit card and blah blah blah blah blah so maybe you could talk through like how Chargmate is solving that and yeah some of the facts and figures behind that

Brad Crist (00:04:36): Absolutely. So what we found was contrary to how the media might portray the problem of a lot of physical damage, that really only made up a small segment of the issues. It was really about user error, payment problems, app issues, glitches between between the handshake of the vehicle and the charger and sometimes like occasional faults like in the equipment or loss of power or connectivity. So what we built is first a system that would solve for the driver facing issues that we thought were unaddressed. And this came from our own experience getting stuck at a public charger, spending 10 minutes on hold with a 1-800 number large networks and it reminded us that this is not only a lousy experience for a customer whose first help is a 1-800 number but that there wasn’t really much visibility that the network had either because their call centers were often offshore third parties that were divorced from their operating system so a big part of what we do is using chat and voice agents to handle the customer facing side on the back end integrate into what’s considered like the network operation center or the charge point management system. And a couple ways that then we’ll check the status of the charger. Is this online? Do we see any faults? Are the charger showing available? What are we seeing now in the transaction processing? And that also allows us to take some commands so we can remotely reboot a unit in some cases. We can force like a plug to be unstuck, which can be a tricky issue. So we’re taking on more advanced automation, basically running the asset on the ground through remote commands and AI.

Brian Bell (00:06:11): So this is interesting. So it’s almost like this AI powered SaaS customer service platform, if you will. And you’re kind of solving it in a vertical way for this very particular use case. Why can’t the incumbents just handle this on their own? Like that would be like the main investor question. It’s like, okay, cool. But isn’t that like their job?

Brad Crist (00:06:30): Good question. So there’s a focus right now in the industry of deploying new infrastructure. And we have seen that then operating that making sure it’s reliable and a reliable seamless customer experience is usually a lower priority and some of that came from reliance on grants and trying to win the kind of land grab and plant a flag where you have ownership of the property or lease agreement for a charger in a parking lot, you know, high traffic locations. And what that’s led to then is these teams are really set up to be acquiring new locations, real estate and infrastructure businesses. They’re not really classic So that’s been one challenge is like even if they have the software team, there’s a lot of existing infrastructure to roll out to maintain. The other side of that is like we’ve seen some companies will build their own AI tools, but They tend to be a little bit narrowly focused on just what one operator sees. Whereas what we’ve observed is there might be issues between vehicles and chargers that that single operator hasn’t even seen yet. So there’s an advantage of for us being able are able to set hardware and software agnostic and start to really see the anomalies that exist between different makes models and and asset types. Then the other thing I’d say is like most companies are not in the business of like running customer experience or a call center function. These are typically set up to be kind of BPO. So now we’re coming in and combining our AI with a human call center partner from Ford, Percepta, where we can actually add the AI and human service end to end. So starting to look more of like an AI enabled BPO option as well.

Brian Bell (00:08:10): Very common in business now, especially over the last few years since ChatGPT came out, is these kind of hybrid service models where it’s almost like all businesses now are kind of these AI-enabled agency business, agent businesses that it’s like, hey, why... I have a business and I have this like repeatable problem and I would typically hire somebody and some call center and wherever to deal with this and now you can onshore a lot of that through AI and deflect I mean what’s your deflection rate on like the human deflection rate I guess would be the question

Brad Crist (00:08:40): From calls, we’ve seen about 50 to 70% resolved by the AI. And that is higher when we can also kind of manage some of the policies. But in cases where, hey, any refund should escalate to human or how do we handle sort of the customer’s choice of channel.

Brian Bell (00:08:58): And AI is as bad as it’s ever going to be. And it gets, you know, 10x better, 5x better, depending on, you know, the intelligence per token per unit of whatever. Right. Every year. And so that number is just going to creep up and up and up and up over time.

Brad Crist (00:09:11): What was that first like aha moment?

Brian Bell (00:09:12): Like we got something. I know you come from industry, right? So you knew the problem was hair on fire. But like, what was that turning point in ChargeMates history where you’re like, wow, this is going to work?

Brad Crist (00:09:21): Great question. I think the first feeling of like relief was we had gone through a bunch of interviews with Charge Executives.

Brian Bell (00:09:28): It is relief, isn’t it?

Brad Crist (00:09:29): it is yeah the first the first like contract was like I cried I was like oh my gosh this is like got a contract is real right exactly so so we were actually working with a large publicly traded network who’s a design partner of ours and now we’re finally talking scale so I can’t quite mention yet who they are but I think hearing enough people say like light up or okay I get it you’re gonna kind of fill this gap that we haven’t even been able to measure or our team is kind of chasing their tail running yeah

Brian Bell (00:09:57): we’re spending so many like they’re like what percentage of their revenue are they spending on customer service I mean it’s got to be significant right super high

Brad Crist (00:10:05): when we look at just OpEx it’s like 80 to 90 percent for some of these larger operators

Brian Bell (00:10:10): That’s crazy and that that flows through and the cost of kilowatt hour right they have to build that into the cost right and you know they can’t just charge you whatever they’re getting off the grid they have to like throw in a bunch of margin to pay for all this customer service and maintenance and so you’re you’re basically taking a huge percentage and contributing back gross margin saving the money saving them time solving a problem creating better customer experiences so obvious thank

Brad Crist (00:10:34): you thank you we think it’s inevitable too

Brian Bell (00:10:36): Yeah, that’s why we invested, if anyone’s wondering. So what did you get wrong in your early versions that you had to pivot away from or reiterate on?

Brad Crist (00:10:45): So the initial vision that we started working on part-time was ways for EV drivers. so partly inspired by the road trip experience you should be able to find an available working charger and how great would it be if that could co-locate with something desirable coffee shop clean bathroom you know you there’s so many options of places is to stop with gas stations where you don’t have the same kind of convenience yet with charging. And what we found there was like, while it’s an important problem of navigating and making the like discovering and kind of education easier, there was really hard I think replacing a Google Maps seeing some way that we would be an endpoint you know in their data set since they have such massive distribution and then realizing that the path to like have any kind of data in the vehicle was also a challenge as a third party like the Auto companies are very guarded about that kind of relationship with their driver and they should be. But what we didn’t see was anyone really helping the driver that feels stranded at this charger that the auto company doesn’t control, that the charging network is dealing with so many fragmented vendors in their own ecosystem as well. Our first chat based product. So we piloted with just putting rogue QR code stickers that fed us some survey data. And then we said, what if we could just start helping people generically with advice? Could we start to get specific to like the equipment? and we needed to really know something about the hardware and the asset. And you’re really only effective if you can also see the status of that charger. So I think the early prototypes were very good at giving us like what are the issues people face, but it was taking down the problem. It wasn’t necessarily resolving it. We might deflect or learn, which was really valuable insights, but there’s only so much you could do without really seeing what is chargers this user in front of. What are we seeing in terms of its status? Do we have some ability to write commands or update a work order log? So I think that was the big thing is realizing this is much more than just a chat bot. This has to be a deeply integrated system.

Brian Bell (00:12:50): Let’s talk about, you know, why doesn’t, you know, an existing, I don’t know, like Zendesk or Intercom or something like that combined with some AI stuff, just replicate this quickly and easily.

Brad Crist (00:13:01): Yeah, you know, it’s something we thought a lot about. And I think Intercom is focused on a more kind of B2B commerce focus on Desk obviously has their support platform. No one’s really knowledgeable of the systems, the intricacies of like integrating software with with firmware and hardware. To give you an idea, like one of our clients has 45 hardware products that they have to manage, different SKUs. One of those SKUs has 722 unique error codes. So really challenging for anyone human to grapple with. but a good fit for AI that’s thoughtfully designed and I think that’s the other thing is like some people will decide to build on their own right this is kind of a big question of build or buy and now that AI and software is easy and plentiful but our bet is that you know it is still expensive to build quality products and maintain them. And so as new vehicles are getting the market, as there’s new hardware and new systems, that’s exactly what we’re keeping up with. And so we see some network effects of like the more vehicles, the more hardware, the more resolution rates, that are using the more failure modes we can solve, the higher the resolution rate, the more likely we’ll be able to serve customers. So yeah, I think it’s a bet that diversity of data, the kind of need to integrate deeply of the systems and workflows. And the other component is like Like we’ve been thoughtful about tying into human teams. I mentioned the call center partner also now working with some field service folks around automating work orders, kind of matching driver feedback against backend telemetry. And I think that human reinforcement loop is really what drives a moat as well for anyone that’s like working in an AI, a vertical AI business.

Brian Bell (00:14:35): Yeah, and I think the EV charging system and infrastructure is as simple as it’s ever going to be right now. We’re just going to have more vehicles, more protocols, more vendors. I mean, maybe over time it kind of, you got, you know, some of the big ones, but yeah, there’s going to be a complexity going forward that ChargeMate’s uniquely suited to kind of handle, right? I hear that the industry’s moving over to the Tesla version of plugs. Like Rivian’s already starting to do this and maybe Lucid as well. They are.

Brad Crist (00:15:03): Yeah. Yeah.

Brian Bell (00:15:04): I had one I had one of those adapters I have one in my Tesla as well just in case I’m somewhere where I can’t use the Tesla charger and I’m very familiar with having to use that so what are you excited about in the future you know look out a year five years ten years kind of taken three guys the industry

Brad Crist (00:15:18): She’s really excited about autonomous cars. I think that’s something that that those of us have been working in energy and electric vehicles realized quickly is that this ushers in, you know, connected vehicles, autonomy and sharing. So I think we’re going to see electric And so that’s exciting from just a big industry tech shift perspective. It’s safer, but it also means tremendous growth in infrastructure. and I don’t think we’re at a point yet where these are going to be robotically charged there’s still very much a coordination problem with humans and depots but there’s also a lot of speculation about like new wireless charging technology much

Brian Bell (00:16:01): the way we can kind of charge a cell phone you drive over the plane and wirelessly exactly charges yeah what was that Israeli was it Israeli company Better Place or what was it called that had the swappable battery packs and raised like 500 million dollars yeah I remember that

Brad Crist (00:16:19): There’s a few few battery swapping technologies that have took off in China but we

Brian Bell (00:16:24): haven’t had the scale really here in North America yeah we’re becoming a Europe in a way in some cases what else what else over the next five or ten years are you excited about

Brad Crist (00:16:34): Yeah I think our bet is that humans are going to interact with machines increasingly through natural language right through chat and voice and so I think our big picture view is that physical infrastructure will be self healing will be capable of communicating in a more natural And I think energy has historically felt like something we don’t think about, or it’s confusing, and being able to make that like a really simple, understandable concept for people. I think what I’m excited about, at least in our industry, is that these charging operators now under huge pressure to be profitable and as you mentioned they’re selling a commodity with like a low margin electricity so that is a lean team that’s like automated and AI driven and so we really are part of helping companies move to profitable with new automated tools so I think there’s kind of both ways what’s happening within these companies to manage lots of assets with a smaller, leaner team and prove that that can be a profitable, thriving business. And there’s kind of the broader just like interaction with physical infrastructure and AI that gets us excited.

Brian Bell (00:17:36): That’s awesome. Well, let’s wrap up with some wrap-up questions. What’s a decision early on that almost killed the company? Hmm, that’s a great question.

Brad Crist (00:17:44): I was pretty close to building out like this kind of Waze-like route planning feature with... some fairly expensive like contractors. And I realized quickly like, no, this is not something I can do as a single founder. Like I need a partner that is equally invested. So I was really grateful to meet Brian Lang, my CTO, who has a background in natural language processing. and had a really strong interest in energy and had worked in automotive as well. So one was finding a great co-founder, but I think the other was finding a paying customer, like a design partner pretty quickly. And the move from a B2C product where we felt like we’re going to spend tens of thousands of dollars to build and acquire users with a hope of monetizing later, instead going B2B and solving for a pain that The operators paying $10 to $15 a call. Let’s start there and then find all the other kind of related costs in operations, support, maintenance that now we can expand to. So I think finding the right painful wedge and building from there was a good learning.

Brian Bell (00:18:44): What’s a belief you have about EV infrastructure that most people would disagree with?

Brad Crist (00:18:49): I mean this is becoming clearer now but like that there’s it can it can be really simple and easy for folks that can charge at home that have public charging and so I have to be careful how much I point out hey here’s problems with charging because I don’t want to scare my own friends and family away right but like there’s a some growing pains so I think that’s something that’s constantly on my mind what’s the

Brian Bell (00:19:09): single hardest lesson you’ve learned selling into this market

Brad Crist (00:19:13): Candidly, I think sales cycles are challenging with larger enterprises and like going to some of the larger operators where we knew we would have scale and be able to make an impact, we quickly realized was slower to get into a roadmap to feel like this would be a scalable deployment. So that meant for us like proof points with smaller enterprises, smaller growing charging companies that were less well known. But it also meant like making the product really easy to get started, finding easier ways to integrate with backend systems, and then making it easy to scale. Now that we are live with voice, we can take on the call center and that’s sort of the first interaction point with the end user. And then it makes some of the scale up smoother for us.

Brian Bell (00:19:55): So if you had to start over today, what would you do differently in the first six months?

Brad Crist (00:20:00): I think we would and this was a challenge with like the AI catching up but like we made the decision to work with chat as an interaction point but it’s really voice and like the kind of natural disposition to like pick up a phone and call and so maybe that would have been where we started first was like focus on kind of the like verbal interaction yeah I think that’s probably the right answer makes sense I

Brian Bell (00:20:22): mean if you think about the user experience too working your way back from how are people going to react when the charging things not working they’re not going to just sit there in their car and chat they’re gonna call they’re like what is going on what’s a what’s a metric uh you pay attention to that’s kind of the most important for your business so

Brad Crist (00:20:38): resolution rate like how many issues do we fix without following a human the you know natural engagement rate of like how are people interacting with AI relative to to like a human call center and I think contrary to popular belief like we don’t just want to to talk to a human in a customer technical support capability. We want to solve our problem. So we’ve seen engagement rates sometimes two or three times higher with the automated channels compared to waiting for a human operator, which has been a good insight. People do engage, especially if it’s effective. So yeah, the engagement resolution rate. And then we’re really focused on proving, hey, this is recovering lost revenue. So for one of our clients, we saw lift in their success rate by about five and a half percent. When we extrapolate that across like a larger network at scale, that’s a couple million dollars in margin, which for these companies is really meaningful. So now just like finding the right opportunity to measure that at scale and then also measure improvement and satisfaction retention for the driver.

Brian Bell (00:21:36): Yeah, that makes sense. So if reliability, if you solve reliability and charging, right, or at least that customer service is taken care of, where is the value shift next? Where should investor be investing is another way of asking that question.

Brad Crist (00:21:49): Yeah, I think there’s a lot to automate the enterprise, like beyond these kind of our vertical wedges support, we’re now seeing opportunities to get into automating sales, installation, commissioning, as well as some of the operations and maintenance. So I think like looking at broader functions and capabilities that that can be automated where AI moves next beyond like call centers and customer support being the low hanging fruit and I think in our industry seeing AI as this sort of like integration layer operating layer that can help bridge gaps between disparate systems and what does that look like now when you’re interacting with lots of different assets that’s that’s pretty interesting to us nice any final thoughts on

Brian Bell (00:22:30): charge mate before I let you go

Brad Crist (00:22:32): Yeah, thank you so much for your support from Team Ignite Ventures. We’re still raising, so would love the opportunity to meet other investors. I think, you know, we’ve talked a lot about EV charging, but we see this as a larger opportunity. We have clients that would ask if our AI can also manage not just chargers but health of the network the battery or building management system other on-site assets so we see this as the first kind of wedge and first beachhead segment by a much larger opportunity for AI to provide a better customer experience and reliability for all sorts of energy assets yeah and where

Brian Bell (00:23:07): Where can folks interested in both customers and investors find you online?

Brad Crist (00:23:11): Were at chargemate.ai and on LinkedIn Bradford Crist

Brian Bell (00:23:16): All right Brad well thanks so muchfor coming on. Really enjoyed it.

Brad Crist (00:23:22): Thanks, Brian.

He was raising money for an AI company, and the pitch was good. Maybe too good. The product was working, customers were interested, the demo had that little electric feeling every investor wants to feel before writing a check. Then he got to the market slide and said, almost casually, “If models keep improving at this pace, this becomes a labor market replacement.”

He meant it as upside.

Everyone in the room understood the sentence in two ways at once.

As a startup pitch, it meant the company might be huge. As a description of the economy, it sounded like a grenade rolling under the table.

That is the strange mood right now. AI is being discussed like software, priced like infrastructure, funded like a gold rush, and feared like a constitutional crisis. One minute the conversation is about revenue per customer. The next it is about whether work, wages, interest rates, sovereign debt, and money itself are about to be pulled into a different gravitational field.

People feel the ground moving, then reach for the biggest available words. Singularity. Collapse. Bubble. End of capitalism.

The end of capitalism is probably the wrong frame. Capitalism has a long history of swallowing its supposed replacements and issuing them a subscription business model. AI fits that pattern for now. The first winners are chip companies, cloud providers, model labs, data-center operators, energy suppliers, software incumbents, and firms with distribution. In plain English, capital is winning.

The sharper question is whether the version of capitalism we have been living inside can survive the arrival of machine intelligence without rewriting its financial plumbing.

Because the old plumbing is already under pressure.

For forty years, the rich world got used to falling rates. That sounds technical. It was everything.

Falling rates made houses more expensive and monthly payments feel manageable. They made stocks more valuable because future profits were discounted at cheaper prices. They let governments borrow without voters feeling the cost. They made venture capital feel rational even when startups burned money for years. They allowed private equity to buy companies with debt and call the arithmetic strategy. They trained everyone to believe the future could be pulled into the present and financed.

Then inflation came back.

It did not return as a theory. It returned as rent, groceries, insurance, energy bills, wages, mortgage payments, and government interest expense. Suddenly money had a price again. Time had a price again. The future stopped being cheap.

That is why this moment feels larger than another rate cycle. AI is arriving at the same time the debt machine is losing its lubricant.

The uncomfortable part is that both forces point in opposite directions. AI promises abundance. The bond market is pricing scarcity. AI says output can explode. Debt says tomorrow has already been spent. AI says labor may become optional in more parts of the economy. Governments still tax wages, payrolls, income, consumption, and corporate profits as if the industrial economy remains the template.

The story gets messy fast. Good. The clean versions are usually wrong.

The low-rate world made everyone a little delusional

Low interest rates do something to judgment. They make distant possibilities feel close.

A dollar ten years from now is worth more today when rates are low. That single fact shaped the whole financial culture of the last decade. Growth stocks, venture funds, crypto tokens, real estate, private credit, and government deficits all benefited from the same background condition. If the cost of waiting is low, dreams get expensive.

Startup investing became one of the purest expressions of that world. A seed company with no profit could raise at a high valuation because the market was willing to pay for a story that might become real five or ten years later. The story did not need to be fake. Many were real. The issue was timing. Low rates compressed the distance between promise and proof.

Governments behaved in a similar way. They did not pitch investors with product demos, but they sold the same basic idea. Lend us money now. Growth, credibility, and future tax receipts will take care of the rest.

For a long time, markets accepted the pitch.

That pitch becomes harder when rates rise.

A government with high debt can function when interest costs are low. It can even look prudent for a while. The danger shows up when old debt rolls into a higher-rate world. The country does not need to default for the pressure to become real. Interest payments crowd out other spending. Voters dislike tax increases. Politicians dislike spending cuts. Central banks dislike inflation. Bond markets dislike fiscal denial.

Everyone wants someone else to absorb the pain.

This is where Japan matters.

Japan spent decades as the example people waved around when they wanted to prove that large sovereign debt could be sustained. They had a point. Japan carried enormous debt, weak demographics, and low growth without the clean collapse predicted by simple textbooks. Domestic savings, institutional bond ownership, low inflation, and central-bank credibility gave the system room.

Now Japan is more interesting as a warning label. Once an economy builds itself around ultra-low rates, leaving that world becomes dangerous. Higher rates can support currency credibility and fight inflation, but they also increase the state’s debt-service burden. The central bank becomes trapped between price stability and fiscal arithmetic.

That trap is not uniquely Japanese. Japan just got there earlier.

The United States has more growth, better demographics, deeper capital markets, and the global reserve currency. Those advantages are enormous. They are also easy to abuse. A reserve currency is a privilege until politicians start treating it like a magic trick.

The old assumption was that developed-market sovereign debt sat at the bottom of the financial system as the safest asset. Increasingly, it looks more like the central political asset. The risk is less about formal default and more about repayment through inflation, financial repression, taxation, currency weakness, or pressure on domestic institutions to absorb government debt at unattractive real returns.

That sounds abstract. It means savers pay. Sometimes slowly. Sometimes without a vote.

AI enters the room at the worst and best possible time

If you wanted to design a technology that could rescue indebted economies, you would design something like AI.

Debt becomes easier to handle when real growth rises. More output means more income. More income means more tax revenue. More tax revenue makes interest costs less painful. If AI lifts productivity enough, a lot of terrifying debt charts start to look less terrifying.

That is the optimistic case. Intelligence has always been the scarcest input. Make intelligence cheap and everything changes. Drug discovery accelerates. Software gets easier to build. Education improves. Customer support becomes cheaper. Manufacturing gets more automated. Scientific research speeds up. Small teams do work that once required entire departments. Governments deliver services with less waste. Healthcare gets less broken. Maybe.

The word “maybe” is carrying a lot of weight.

AI has to pass through the real economy before it saves the fiscal state. The real economy is full of stubborn objects. Power grids. Hospitals. Procurement departments. Lawsuits. Unions. Insurance codes. School boards. Permits. Data silos. Legacy software. Human fear. Human vanity. Human incompetence in high places.

Anyone who has watched a large company buy software knows the problem. The demo takes twelve minutes. The deployment takes twelve months. Sometimes longer. Then half the team refuses to use it because the old spreadsheet still feels safer.

This is why the AI boom can be real and overhyped at the same time. That combination is common in major technology shifts. The technology changes the world. The first wave of financial claims on that change still gets too excited.

The internet was real. The dot-com bubble still burst.

AI may follow a similar pattern, with larger numbers and more political heat. The infrastructure buildout comes first. Chips, cloud capacity, data centers, grid connections, cooling systems, software tooling, security layers, enterprise pilots. Money pours into the rails before anyone knows the final shape of the train schedule.

The early AI economy is less like magic abundance and more like a massive construction project with better marketing.

That matters for inflation and rates. People talk about AI as if it will make everything cheaper. Over time, it might. In the early phase, it demands scarce inputs. Advanced chips. Electricity. Land. Engineers. Transformers. Fiber. Water. Specialized construction. Political permission.

Scarce inputs do not become cheaper because a slide says “exponential.”

So the transition can be inflationary before it becomes deflationary. Companies spend heavily. Utilities strain. Supply chains tighten. Workers with relevant skills become expensive. Asset prices rise because investors believe the prize is huge.

Then everyone looks around and asks why the abundance feels so costly.

The bubble is built into the structure

AI is perfect bubble material because the upside is huge and hard to bound.

A normal business has a market. AI keeps trying to become the market underneath other markets. Software, services, education, law, medicine, finance, advertising, entertainment, coding, sales, operations. Every category can be reimagined if machine intelligence gets cheap enough and reliable enough.

Investors hate missing platforms. They especially hate missing platforms after seeing the last generation of platform companies become some of the most valuable firms in history. That memory changes behavior. It makes capital faster, more anxious, and more forgiving of vague answers.

The most dangerous sentence in a financing meeting is “this could be the interface for all work.” It may be true. It may also justify almost any valuation if said with enough conviction and accompanied by a clean demo.

The AI market has another bubble ingredient: reflexivity.

A company valued highly can raise more money. More money buys more compute and talent. More compute and talent improve the product. A better product attracts more users. More users improve distribution and data. The improved story raises the valuation again.

For a while, the financial loop and the product loop reinforce each other. It feels like destiny from the inside.

Then the boring questions arrive.

How much will customers pay?
How much gross margin survives when inference costs are included?
Who owns the workflow?
Can open-source models compress pricing?
Does the product reduce headcount or just create another software line item?
Will enterprises trust it with sensitive decisions?
What happens when every competitor has similar model access?

Experienced investors ask these questions early because they have seen the movie where a product is impressive and the business is mediocre. Intelligent newcomers often underestimate this distinction. A demo can be magical while the profit pool remains slippery.

AI makes that distinction harder because the demos are so good. The machine talks back. It writes code. It makes images. It drafts contracts. It answers customer tickets. It feels like labor on a screen.

But a company does not get paid for sounding like the future. It gets paid for controlling something valuable that customers cannot easily get elsewhere.

That control might come from distribution, proprietary data, deep workflow integration, regulatory trust, brand, switching costs, hardware access, or ownership of the customer relationship. Without control, AI features become table stakes. Table stakes are useful. They do not always create great businesses.

This is one reason AI can inflate public markets and disappoint private investors at the same time. The biggest incumbents may capture much of the value because they already own the customers, infrastructure, and balance sheets. Startups create the heat. Platforms collect the rent.

The labor problem is the real problem

People talk about money supply because it is visible. Central-bank balance sheets have charts. Interest rates have announcements. Debt has numbers with commas.

The deeper issue is income.

Modern capitalism works because most people get purchasing power by selling labor. They work, earn wages, buy goods and services, pay taxes, and support the state. The system has plenty of unfairness, but the loop is understandable.

AI threatens to weaken that loop.

The first affected workers are not only factory workers or call-center workers. They include programmers, analysts, designers, marketers, lawyers, consultants, support reps, junior bankers, copywriters, recruiters, accountants, tutors, and operations managers. The polite phrase is “productivity enhancement.” The less polite version is that many white-collar tasks are being repriced.

The job does not need to disappear for wages to come under pressure. If one person with AI can do the work of three people, the labor market changes. If junior tasks get automated, the training ladder changes. If software becomes easier to build, some engineering work becomes more valuable and some becomes more commoditized. If every company can produce more content, content gets cheaper. If every analyst has a tireless assistant, basic analysis loses scarcity.

This is where the social contract starts to creak.

An economy can grow while many workers feel poorer. Asset owners can become richer while wage earners feel replaceable. Stock indices can rise while households feel trapped by rent, insurance, childcare, healthcare, and debt.

That combination is politically explosive because people do not experience GDP. They experience cash flow.

If AI raises output while reducing labor’s share of income, governments will need new ways to distribute purchasing power. That does not require a utopian ideology. It follows from arithmetic. If wages become a weaker channel for mass income, some other channel has to carry more weight.

That could mean larger earned-income credits, wage subsidies, public services, universal basic income, sovereign wealth funds, public stakes in AI infrastructure, heavier taxes on monopoly profits, taxes on capital gains, or new forms of citizen dividends.

The exact policy will vary by country. The direction is harder to avoid. A labor-light economy needs a broader income-distribution mechanism or it becomes socially unstable.

This is where the “end of capitalism” conversation gets confused. AI does not automatically end private ownership. In the early phase, it strengthens private ownership. The owners of compute, energy, distribution, models, and capital gain more leverage.

The pressure comes later, when voters ask why the machine produces abundance and their lives still feel financially narrow.

GDP may start lying more than usual

Gross domestic product measures market activity. It is useful. It is also clumsy.

AI will make it clumsier.

Suppose an AI tutor gives every child access to personalized instruction at near-zero cost. Human welfare could rise enormously. Measured GDP might barely notice if the service is cheap or bundled into a subscription.

Now suppose companies spend hundreds of billions on data centers, chips, consulting, and enterprise software, then use much of it to generate ads, spam, synthetic media, and internal slide decks nobody reads. GDP rises. Human welfare may not rise much.

Both outcomes can happen at once.

This matters because debt lives in the measured economy. Governments collect taxes from transactions, wages, profits, capital gains, property, and consumption. Bond markets care about nominal growth, inflation, and fiscal receipts. They do not give a government full credit for consumer surplus that never turns into taxable cash flow.

A free AI doctor would be socially extraordinary. It helps the state’s finances only if it reduces public healthcare costs, creates taxable income, or replaces expensive labor in a way the government can capture.

That last phrase is ugly because the topic is ugly. Fiscal systems need claims on value. If value migrates into cheap digital abundance, untaxed capital gains, offshore profits, open-source models, or consumer surplus, the state can become fiscally stressed while citizens are surrounded by powerful technology.

A country can be technologically rich and fiscally weak.

That is one of the strange possibilities ahead.

Money will still matter

The dream version of the singularity says money fades because abundance wins. Maybe in the distant limit. Before that, money may matter more because claims on the future will become more contested.

Money is a claim on output. Debt is a claim on future money. Equity is a claim on future profits. Wages are claims earned by selling labor. Government bonds are claims on future taxpayers. If AI changes production, every claim built on top of production starts to move.

That is why the topic feels so large. It is not just about better chatbots or faster coding. It is about who gets paid when intelligence becomes software.

Money supply matters, but the next phase may depend even more on collateral. Modern credit systems lend against assets. When AI inflates the market value of certain assets, it expands the collateral base. Companies can raise more capital, borrow more cheaply, pay employees in stock, acquire competitors, and fund larger infrastructure plans.

The asset price becomes part of the engine.

This can work beautifully while confidence holds. Rising valuations fund real investment. Real investment improves products. Better products support the original valuations. Everyone feels brilliant.

If expectations break, the loop runs in reverse. Equity falls. Credit tightens. Capex slows. Tax receipts weaken. Governments respond with support. Central banks feel pressure. The private bubble becomes a public problem.

This is the part people tend to miss. Large asset bubbles rarely stay private. Once enough jobs, tax receipts, pensions, retirement accounts, and infrastructure plans depend on inflated values, the state gets dragged in.

AI could produce the largest version of that story because the investment needs are enormous and the promised prize is civilizational.

Interest rates could go either way

The simple view says AI means deflation, so rates should fall.

That may be right eventually. It is a weak guide for the transition.

In the near term, AI can push rates higher by increasing investment demand. Everyone wants compute. Governments want national AI capacity. Militaries want autonomy and intelligence systems. Companies want automation. Cloud providers want data centers. Data centers want power. Power wants grids, permitting, turbines, substations, transformers, and fuel.

That is capital demand. Capital demand can push real rates up.

AI can also create inflation through wealth effects. If asset owners get richer, they spend more. If companies invest aggressively, demand rises. If electricity and commodities become bottlenecks, prices rise. If governments subsidize domestic AI infrastructure, deficits widen.

Over time, AI may push in the other direction. Better software lowers costs. Better science improves energy. Better robotics reduce manufacturing costs. Better logistics reduce waste. Better education raises human capability. Better medicine reduces expensive failure.

So the path could be high-pressure first, disinflation later.

Markets are tempted to price the later world before the current one has paid for the buildout. That is how long-duration bubbles form. Investors discount the promised abundance while the economy is still buying the servers.

Sovereign debt becomes a bet on productivity

Government debt is usually discussed as a moral drama. Too much spending. Too little discipline. Irresponsible politicians. Entitled voters.

Some of that is true. It is also incomplete.

Debt sustainability depends on growth, rates, inflation, and political capacity. A country can carry high debt if growth is strong, rates are manageable, and investors trust the system. A country can struggle with lower debt if credibility breaks.

AI enters as the great productivity bet.

If AI meaningfully raises real growth, governments get relief. More income. More profits. More taxable activity. Lower unit costs. Better services. Maybe smaller deficits.

If AI disappoints, debt math becomes harsher. If AI concentrates gains in companies and individuals that avoid taxation, governments may receive less relief than the productivity headlines suggest. If AI disrupts labor income faster than fiscal systems adapt, deficits can widen even as technology improves.

That is the dangerous middle outcome. The economy gets smarter. The state gets poorer. Politics gets angrier.

The fiscal question becomes less about whether AI creates value and more about whether public institutions can capture enough of that value to fund obligations.

The United States has a particular advantage here because many leading AI companies, chip designers, cloud platforms, and capital markets sit within its orbit. That gives the U.S. a better chance of taxing, regulating, and benefiting from the boom. It also creates complacency. Dominant countries often mistake structural advantages for permanent exemption from arithmetic.

Arithmetic has a long memory.

The new bottlenecks

Every age has its bottlenecks.

In the industrial age, machinery, oil, labor, factories, ports, and finance mattered. In the software age, talent, code, distribution, and network effects mattered. In the AI age, the bottlenecks look different.

Compute matters.
Energy matters.
Data matters.
Distribution matters.
Trust matters.
Capital matters.
Regulatory permission matters.
Ownership matters.

That last one may matter most.

If AI becomes a general-purpose production engine, then owning the engine becomes the central economic fact. A person with no ownership may get cheaper services and worse bargaining power. A person with ownership may get compounding claims on machine output.

This is why the singularity debate can sound mystical while the practical question is brutally simple. Who owns the productive assets?

If ownership stays concentrated, AI becomes a machine for widening the gap between capital and labor. If ownership broadens through public markets, pensions, sovereign funds, employee equity, public investment vehicles, or tax-and-transfer systems, AI can support a wider prosperity.

The technology does not decide that. Institutions do.

And institutions are slow. Slower than model releases. Slower than markets. Slower than fear.

What experienced investors are watching

The public conversation loves dramatic questions. Will AI take all jobs? Will money disappear? Will capitalism end?

Experienced investors tend to ask more grounded questions because grounded questions reveal where the money goes.

They ask whether customers are paying more because the product is useful or because the budget cycle is full of AI experiments.

They ask whether the company owns the customer relationship or sits as a feature inside someone else’s platform.

They ask whether gross margins improve with scale or get eaten by compute costs.

They ask whether the product replaces labor, increases labor productivity, or creates more work under a shinier label.

They ask whether the company has proprietary data that matters.

They ask whether the buyer can measure return on investment without pretending.

They ask whether a model upgrade from OpenAI, Anthropic, Google, Meta, or an open-source project can erase the company’s advantage.

They ask whether the founder is describing a business or narrating a future.

That last distinction matters. A future can be correct while a company inside it fails.

The same discipline applies at the macro level. AI can transform the economy while many AI investments lose money. AI can raise GDP while worsening inequality. AI can reduce prices in some categories while increasing pressure on energy and capital markets. AI can help sovereign debt over time while worsening deficits during the buildout.

The world can get richer and more unstable at the same time.

The end state is allocation

The more I think about the singularity economy, the less I think the central question is production.

Production is the fun part. More intelligence means more discovery, more automation, more optimization, more software, more science. It is easy to get excited about that because the upside is real.

The harder question is allocation.

Who gets the output?
Who gets the income?
Who pays the debts?
Who owns the machines?
Who taxes the rents?
Who absorbs workers whose tasks have been repriced?
Who controls the compute?
Who decides whether abundance arrives as lower prices, higher profits, larger transfers, or stronger monopolies?

Those questions sound political because they are. Every economic regime eventually becomes a political settlement. The AI regime will be no different.

Money will not vanish early in that process. It will become the scoreboard for a more intense fight over claims. Wages, profits, taxes, benefits, debt payments, capital gains, subsidies, and inflation will all become ways of deciding who receives the gains from machine intelligence.

The cleanest positive scenario looks like this. AI raises productivity. Growth improves. Governments capture enough tax revenue to manage debt. Competition and open-source models prevent permanent monopoly rents. Prices fall in important categories. Workers use AI to become more productive rather than broadly replaceable. Public policy updates the income system before social trust breaks. Energy supply expands fast enough to support the compute buildout.

That scenario is possible.

The darker scenario is just as easy to describe. AI raises asset prices faster than real productivity. Infrastructure spending booms. Labor income weakens. Sovereign debt costs keep rising. Governments become dependent on bubbly tax receipts. Incumbents capture the profit pools. Central banks face inflation they cannot cleanly fight. Voters grow hostile toward a system that produces technological miracles and household anxiety in the same year.

That scenario is also possible.

The likely path has pieces of both. A boom, a bust, real productivity, wasted capital, monopoly profits, open-source deflation, better tools, worse politics, sovereign intervention, and a long argument over ownership.

Messy. Human. Expensive.

What this moment really is

The singularity will not first arrive as a glowing line on a chart. It will arrive through budgets.

A cloud provider signs another power agreement. A government pays more interest on old debt. A startup raises money at a valuation that assumes a labor market will bend. A household refinances nothing because the rate is too high. A company freezes hiring because its AI tools are good enough. A central bank holds rates while politicians complain. A pension fund buys the index because it cannot afford to miss the only growth story large enough to matter.

That is how a new regime starts to feel real. Not as one grand announcement. As a thousand decisions that begin to rhyme.

So no, capitalism is not ending in some clean cinematic sense. The stronger claim is that the low-rate, labor-centered, debt-tolerant version of capitalism is under strain. AI gives it a possible escape route and a possible accelerant for its worst tendencies.

It can grow us out of debt or inflate the biggest asset bubble in history.

It can broaden access to intelligence or concentrate power around the owners of compute and distribution.

It can lower the cost of living or raise the value of assets faster than wages can follow.

It can make governments more capable or expose how poorly their fiscal systems fit a post-labor economy.

The mistake is expecting one answer.

For a while, AI will be productivity miracle, investment mania, fiscal hope, labor shock, and political accelerant all at once. That is why the moment feels so hard to name. We are trying to describe a technological transition using financial language built for a slower world.

The old economy asked how much labor and capital could produce.

The new economy asks what happens when intelligence itself becomes capital.

Once that question is live, everything downstream starts to move. Money. Debt. Rates. Wages. GDP. Asset prices. The state.

The singularity has a balance sheet.

We are only beginning to read it.

The Lean Startup became required reading across Silicon Valley, shaping how founders build products, test markets, and scale companies. But in his new book, Incorruptible: Why Good Companies Go Bad and How Great Companies Stay Great, Ries shifts from product strategy to something much deeper:

Why do successful companies eventually betray the very thing that made them great?

And more importantly:

Can founders stop it from happening?

After speaking with Ries on the Ignite Podcast, one thing became clear: this is not another “culture matters” business book. It’s a direct attack on the modern incentives driving corporate behavior.

His argument is simple but uncomfortable:

Most companies don’t fail because they lose.

They fail because success turns them into targets.

The Core Problem: “Financial Gravity”

Ries introduces a concept he calls financial gravity — the pressure organizations feel to conform to the priorities of whoever controls capital.

At first, companies exist to solve problems.

Then they scale.

Then the incentives shift.

Leadership meetings become dominated by conversations about quarterly expectations, analyst reactions, margin optimization, growth rates, and stock performance. Over time, the mission slowly becomes secondary to financial engineering.

According to Ries, this doesn’t usually happen because founders are evil.

It happens because the system rewards extraction.

The most striking part of the conversation was his claim that many executives don’t even realize the shift is happening while it unfolds. What starts as “just this quarter” becomes permanent institutional behavior.

That observation explains a lot about modern business.

Why products degrade after acquisition.

Why beloved consumer brands suddenly become unusable.

Why tech companies increasingly optimize for ad revenue over user experience.

Why healthcare systems maximize billing efficiency instead of patient outcomes.

Why companies that once felt mission-driven eventually feel hollow.

Ries argues these aren’t isolated failures.

They’re structural outcomes.

Why Costco Matters More Than Most Startups

One of the most compelling stories in the episode involved Sol Price, the founder of FedMart and later Price Club — the precursor to Costco.

Price believed retailers had a fiduciary duty to customers. He capped margins, paid employees well, and even told customers when competitors offered lower prices elsewhere.

That level of customer trust created enormous long-term value.

But investors hated it.

Eventually, Price was forced out of his own company. FedMart collapsed within years after leadership prioritized extraction over trust.

But Price started over.

That second company eventually became Costco.

Ries uses Costco as an example of a company with what he calls structural integrity — a governance system strong enough to resist short-term pressures.

The point is not that Costco is perfect.

The point is that incentives matter more than slogans.

Every company says it cares about customers.

Very few build structures that protect customer value when financial pressure arrives.

The Founder Trap

One of Ries’ most controversial claims is that founders routinely misunderstand what it takes to preserve a company’s ethos.

Most assume culture transfers automatically.

It doesn’t.

A founder can personally value innovation, craftsmanship, or long-term thinking while simultaneously building systems that reward bureaucracy, fear, and short-term optimization.

This explains why so many founders eventually lose control of their own companies — even while technically remaining CEO.

As organizations scale, employees stop responding to the founder’s intentions and start responding to the incentives embedded in the system.

That distinction matters.

Because systems outlive charisma.

Ries argues that governance structures — board composition, voting rights, ownership models, incentive design — shape company behavior far more than motivational speeches or culture decks.

That may sound abstract, but the evidence is hard to ignore.

Many iconic founders were eventually removed from companies they created.

Steve Jobs.

Edwin Land.

Numerous venture-backed CEOs after IPO.

According to Ries, most founders dramatically underestimate how vulnerable they are once external capital gains influence.

AI Makes This More Dangerous

The conversation became especially interesting when discussing AI.

Ries believes AI amplifies existing organizational incentives.

If an organization is extractive, its AI systems will likely become extractive too.

If a company prioritizes trust and long-term alignment, AI can strengthen those advantages.

This is why he frames AI primarily as a governance problem, not just a technology problem.

Consumers and enterprises are increasingly dependent on AI systems that control workflows, data, recommendations, and decisions. That creates enormous asymmetries of power.

The companies that win long-term may not simply have the best models.

They may be the companies people trust most.

That distinction could define the next decade of technology.

A Different Definition of Profit

The most radical idea Ries proposes may also be the simplest:

Profit should mean creating net new human flourishing.

Not merely maximizing extraction.

He argues that many activities modern finance treats as “profitable” actually destroy value once externalities are included — whether through degraded trust, environmental damage, addiction mechanics, monopolistic behavior, or institutional decay.

You don’t have to agree with every part of his framework to recognize the underlying tension:

Modern markets often reward short-term optimization even when it undermines long-term value creation.

That tension increasingly defines tech, media, healthcare, finance, and AI.

And founders are caught in the middle.

Why This Conversation Matters

Startup culture spends enormous time discussing product-market fit, growth loops, fundraising, and scaling.

Far less attention is paid to what happens after success.

Ries is arguing that governance itself may become the defining competitive advantage of the next generation of companies.

Not just better products.

Better structures.

Stronger incentives.

More durable missions.

Companies capable of surviving success without becoming corrupted by it.

Whether you fully buy the thesis or not, the question he raises is difficult to ignore:

If your company became massively successful tomorrow, would its incentives still align with its mission five years later?

Most founders probably assume the answer is yes.

History suggests otherwise.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:

00:01 — Intro: Eric Ries Returns to Ignite
00:40 — Why Good Companies Go Bad
02:16 — Good to Great vs Incorruptible
03:10 — Financial Extraction & Corporate Decline
06:53 — The Long-Term Stock Exchange Experiment
08:33 — Financial Gravity Explained
10:26 — Why Founders Succumb to Short-Term Pressure
14:16 — The Moment Companies Become Corrupted
14:48 — The FedMart & Costco Story
19:35 — Why Markets Reward Extraction
22:47 — Shareholder Primacy vs Mission Primacy
25:04 — Organizations as Emergent Intelligence
28:13 — Ethos vs Company Culture
31:34 — Why Founders Lose Control of Their Companies
36:24 — Governance Mistakes That Destroy Companies
40:10 — Jeff Bezos, Amazon & Long-Term Thinking
43:20 — Leadership, Profit & Human Flourishing
48:32 — Mission-Driven Business Models
49:11 — Does Human Flourishing Break Capitalism?
52:34 — Blueprint for Building Incorruptible Companies

Transcript

Brian Bell (00:01:31):
Hey everyone, welcome back to the Ignite Podcast. Today we’re thrilled to have Eric Ries back on the mic. He is the creator of the Lean Startup Movement and now the author of his upcoming book, Uncorruptible, Why Good Companies Go Bad and How Great Companies Stay Great. Thanks for coming back on, Eric.

Eric Reis (00:01:45):
Oh, it’s my pleasure. Congrats on the show. I love it. Yeah. It’s growing and thriving. Great.

Brian Bell (00:01:49):
Yeah, thanks. It’s always great to have guests back on, especially when they publish books. It’s kind of like my favorite type of podcast. So tell me what this book is about. It’s not really about how to build companies. My understanding is it’s why companies we admire inevitably betray their mission and what it actually takes to prevent that. Maybe you talk to me about what the book’s about and why you wrote it.

Eric Reis (00:02:09):
Yeah I’ve noticed because you know I’m a big believer in feedback as you might have heard so I had 600 people test read the book they generated more than 10,000 comments so I have a really good sense of like how people respond to the book and I was able to really track like I could tell the book was getting better as I was writing it I apologize to all the test readers of the first version it was really grim, sorry. I did my best. Anyway, that’s how you make it better. And I noticed there are basically two kinds of readers for this book, what I call the why readers and the how readers. They’re right there in the subtitle. First of all, there’s people who are like, why does this keep happening? Why are every company I admire eventually turning how come like when private equity takes over my favorite restaurant I can taste it what’s up with that why is this happening and then the how readers are company builders are leaders and board members and founders and even investors who want to know like we don’t need another manifesto about how our economy is messed up we got plenty of those the real question of this book is how how do you build an organization that is incorruptible that people try this shenanigans on it and it doesn’t work because it’s strong enough to resist so yes it is in some level a company building book But now challenging not just the like product building or strategy level topics like in lean startup. Instead, now we’re getting at what I think of as the underlying structural forces that affect how companies behave and the governance and structure. that determine how strong or weak a company is, which I feel like is a way of thinking about companies that many, many people have not yet encountered. But once you see it, you’re not going to be able to unsee it because we talk about companies as great, but I want to ask, they might be great right now, but are they weak or are they strong? It reminds me of the- Strong companies that will endure.

Brian Bell (00:03:44):
Yeah, it reminds me of the first book I read in business school 25 years ago in Business 300 at University of Washington, Seattle. It was Good to Great. It was like the first book we read. So I mean, you’ve probably read that book and studied it. And what is this book kind of contributing to the business literature that kind of either disagrees with Good to Great or kind of builds on some of the philosophy of that book?

Eric Reis (00:04:06):
yeah yeah yeah the the the ellipsis in the subtitle is actually a good to great homage that’s not we don’t we don’t do that anymore but it’s a bit of a throwback just because I wanted to give my my tip of the hat to to Jim Collins yeah yeah so so this is a book about not the companies that fail because they messed up it’s about the companies that fail because they’re successful one of the biggest things I think we don’t teach leaders in our economy today is that the more successful an organization the more valuable.

Eric Reis (00:04:47):
Not just private equity in restaurants, but like when private equity buys nursing homes, they cut nursing staff for returns. Patient mortality rises 10%. There’s a NBR working paper that showed that. When they buy hospitals, they cut clinical staff. ER deaths rise 13%. My favorite is that when people buy a sports team, if they operate it in an extractive way to maximize profit, the research shows that they will have the lowest quality team and the smallest fan attendance. Increasingly, we’re not customers customers, citizens, fans anymore, patients. We’re a resource to be mined. That’s how we’re teaching people to practice business. And what’s so interesting to me about this is if you talk to people now and say, why is this happening? Most people are like, oh, it’s just inevitable. That’s just, that’s capitalism for you. You know, I should talk to young people. That’s capitalism.

Brian Bell (00:05:34):
It was so crazy, right? I mean, the Fortune 500 turnover over the last hundred years. You used to get into the Fortune 500 and you would stay there for a half century.

Eric Reis (00:05:43):
And now the turnover is- The average lifespan of companies has fallen dramatically. The average holding period of stocks has fallen dramatically. And the average tenure of executives has fallen dramatically. So we are in an era of disposable organizations being led by temporary managers on behalf of absentee owners.

Brian Bell (00:06:01):
What are we doing?

Eric Reis (00:06:02):
And then we’re like, gosh, why is trust collapsing? No kidding. But what’s so strange is that these ideas are new. Our grandparents were very allergic to the idea of making money without creating value. In fact, it’s one of the most ancient teachings about economics there are going back to Aristotle and beyond in almost every religious tradition that there are better and worse ways to make The good way is to create net new value in the world and then capture some of it for yourself. Every other way is not just morally suspect, I think is corrupt. Many of the ways that people make money today, whether we’re talking about betting markets or like all kinds of legalized gambling or like taking companies over to destroy or kill them or called killer acquisitions. I could go on and on and on and on. In our grandparents or great grandparents’ time, not only would those things have been seen as corrupt, they would have been illegal, would have been crimes. In the 19th century, just to give you a sense of how far things have changed in just a hundred and some odd years, in the 19th century, if you tried to take over a company that made a specific thing, in those days, companies were not seen as just about maximizing returns to shareholders. Every company had to be incorporated with what’s called a beneficial purpose. You had to say, this is a company designed to make a railroad. And you had to, in your application to form

Eric Reis (00:07:11):
Ignite Insights

Eric Reis (00:07:26):
to buy. But secondly, if you did succeed and take it over and you tried to change the corporate purpose from make a railroad to enrich my shareholders, that would have been considered a crime and the courts would give you the corporate death penalty. They would void your charter because you’ve gone beyond the scope of what is authorized. So this is not some like newfangled idea. In some ways, this is just a restoration of the ideas that have dominated since the creation of the Joint Stock Corporation until very, very recently.

Brian Bell (00:07:55):
And you’ve had a front row seat at this because you’ve been involved in the inception of the long term stock exchange, as I recall. Maybe you could explain like that was trying to get at this problem in a structural way, right?

Eric Reis (00:08:08):
Yes. Short answer to your question is yes. Yes. I have been at this. I’ve had front row seat to this for 15 years now. We sort of came out 15 years ago. I have helped hundreds or thousands of companies be formed. I mean, I’ve lost count at this point how many founders I’ve helped. And I’ve seen this process by which the thing that made the company worth investing in in the first place gets lost goes by the name of mission or purpose in the book I call it ethos the character of a thing can be lost not just due to outside pressure of course you know private equity is a frequent villain but not like sometimes just when a company goes public sometimes just when And the founder loses control, even to his employees. Brian Chesky very famously said he felt like he was losing control of Airbnb, not to his investors. He had dual class control, they couldn’t do anything. But he lost it to his employees because his employees were in the grip of these best practices. And this is something that I think is just, it’s an epidemic in our society. We have financialized everything. So every kind of organization, like not just for profits, but nonprofits, NGOs, unions, journalism, everything is coming under the sway of these, this kind of best practices monoculture. that is about extraction and exploitation rather than about creating value. And yeah, I tried to do something about it. I built this thing called the long-term stock exchange. Still going, making money, trading stocks, listing companies. The first attempt to really change the listing standards by which public companies are judged. And interestingly, we’ve been in the news recently because you may have seen the headline about that the SEC is thinking of eliminating quarterly reporting.

Brian Bell (00:09:31):
I did see that, yeah.

Eric Reis (00:09:32):
That’s the LTSC petition. Wow that that decision making process was initiated by a petition that LTSC filed with the SEC so I really believe in taking matters into our own hands as builders we have to we can’t just sit back and be like oh the economy sucks these rules suck no we have to we have to take it in hand to change them but this book is not so So much about LTSC style change although that is discussed especially in the later chapters it’s primarily about what we who build and lead companies can do today right now there’s not a single technique in the book you have to wait for the revolution okay everything is something you can do today to protect what makes a company special from people who would try to take it from it’s amazing so you and

Brian Bell (00:10:08):
in the book you argue corruption it’s not necessarily about bad people it’s more structural

Eric Reis (00:10:13):
Yeah.

Brian Bell (00:10:13):
Which is a strong claim. So what is that mechanism that forces good companies to go bad?

Eric Reis (00:10:18):
In the book, I call it financial gravity. It’s one of the underlying fundamental forces that acts on organizations. And it’s the psychological pressure that we all feel to conform to the values of those who have more than we do, because we want something from them. So if you ever see a company go public, I talk to a lot of CEOs who are taking companies public as part of my work at LTSE. You ask them, what’s the biggest difference before and after the IPO? They all say the same thing. Everyone’s watching the stock ticker. In product management meetings, all of a sudden people say things like, well, the market might not like it. The word might does a lot of work in these sentences. You see it in startups. VCs might not like it. We might not be able to raise money. You talk to VCs of these bold contrarian VCs on your board. You tell them you want to do something related to mission protection. They’ll be like, I don’t know. I’m okay with it, but other investors might not like it. It’s like, I thought you were a bold contrarian. Now I have to be just like everybody else. Like, what are we doing here? What happens is there’s this ghost. who’s a secret participant in every meeting whispering in your ear, I don’t know, they might not like it. Now that’s different than finding out what investors actually like. The might does a lot of the work here. And if you want to understand the mechanism, this is an evolutionary instinct that we all carry. You cannot turn off any more than you can turn off your eyes dilating in the dark. What happens is, have you ever watched somebody meet a celebrity for the first time or hang out with a billionaire for the first time? It’s happened to me. it is so gross they become obsequious they’re like they can’t help themselves if you ask them afterwards did you intend to act like that they’ll always be like no they’re a little embarrassed about it why were you such an obsequious but well I just I don’t know what’s going on unconsciously is you’re like gosh this super rich

Brian Bell (00:11:53):
person likes my idea yeah that was like me on our first podcast I was like kind of starstruck because I read your book you know almost 20 years ago oh god it was like oh my god I’m interviewing Eric you know

Eric Reis (00:12:04):
Well I had no idea so you hit it you hit it well yeah but but a lot of people really don’t and it’s right it’s like very obvious so what’s happening is like people are unconsciously always calculating what’s going to help me get ahead in the future what’s going to help me raise money get the promotion whatever it is

Brian Bell (00:12:19):
that I need you see it like your boss conforming to the financial short-term goals of the system rather than keeping the mission at heart

Eric Reis (00:12:27):
And we know from the psychology research that consistently enacted actions eventually become internalized as values. So you start out by saying, I believe in quality over everything, but because the values of our financial system tend to preach short term extraction over value creation, over quality over patient health or safety or whatever else. Over time, you start to unconsciously feel like those things are unaffordable luxuries. And you have to, you have no choice but to compromise just this once to get ahead. Jim Senegal, the founder of Costco, calls it like the corporate version of taking heroin.

Eric Reis (00:12:57):
He tells a story that in Costco- Or crack.

Eric Reis (00:12:59):
Yeah, he’s like literally if we raised prices 3% across the board, we would double our net income overnight and nobody would notice. But the problem with that is once you do it once, you got to do it again because that becomes a new baseline against which you’re judged for future growth. Andrew Mason the founder of Groupon told me this unbelievable story when they took Groupon public you know the whole thing about Groupon for those that don’t remember was like a one email a day they have a special deal of the day you know of something you really loved you get the special email you all sign up for the group thing and you get the discount so they went public on the power of one email a day it was a very simple business model very effective extremely fast growing business once they were public people kept coming into his office and they’d say you know what would perform even better than one email a day two emails a day and at first he’d be like no no no our whole thing is one email a day but he they wore him down over time they kept coming with these spreadsheets showing him the ROI and they’d be like let’s just do the experiment let’s just try it MVP whatever people like very easily forget that the point of the experimentation is to

Eric Reis (00:13:58):
Ignite Insightsights

Eric Reis (00:14:18):
what if we did three emails and next thing you know they’re emailing people too often and the business collapsed he felt powerless in the face of this because he didn’t have the tools he needed to defend his ethos and explain to people why we’re doing it this way even though it’s not ROI maximizing this is the danger of this era of business best practices that leave us blind to the things we need to do to maximize our long-term value creation

Brian Bell (00:14:43):
And in the book, this is the moment of corruption inside of a company, right? Where the visionary, the founder succumbs to the pressure of, okay, maybe it’s worth a test or just this quarter so we hit our numbers. Maybe you could walk us through what that actually looks like in the boardroom, in the kind of the e-staff.

Eric Reis (00:15:02):
Yeah so there’s these moments maybe maybe let me illustrate with the story just to kind of bring the whole thing together okay because I think that sometimes these things can go very abstract like corporate purpose and profit and whatever is very abstract let’s get it super concrete let’s tell let me tell you the story of a guy named Saul Price I don’t know how many of your listeners know the story anymore he’s kind of fallen out of the business lexicon but he used to be very famous because he’s the father of modern retail he invented the big box warehouse format of retail but before he became a retailer he was actually a lawyer And he told the story this way. He said, when I was a lawyer, I learned that I had what’s called a fiduciary duty to my customer, to my client, a person who I serve. So we all want that from our- There’s agency, yeah.

Eric Reis (00:15:41):
Yeah, we all want that. Anytime we’re in an agency relationship with somebody, we want them to put their interests before ours. So when he became a retailer, he asked himself, who’s my client? The customer is my client. If I’m a discount retailer, my job is to look out for the well-being, the financials, financial well-being of my customers so he took that really seriously in the new company he created called Fedmark which if you’ve been in a big box retail store recently you will recognize it intimately right no frills big warehouse large quantity limited selections low price he practiced what are called capped margins every item marked up exactly 14% and no more and employees very happy to work there paid above average wages all the stuff you want from a business He was so committed to his fiduciary commitment to customers that when competitors would undercut him on price, he would post signs in his own store saying, hey, don’t buy this product from me. You can get it cheaper down the street. So as a result, customers trusted FedMart and they would drive miles out of their way to shop there because we’ve all had that experience. When you shop with someone you trust, you don’t have to think too hard about it. You trust you’re not being ripped off. That’s the kind of place it is. Everything met there, you know, his personal quality guarantee, everything at a fair price. So this powered FedMart to tremendous expansion. It eventually went public. But as a public company, he felt this gravitational pressure, investors constantly on him to corrupt. he wanted low prices and high wages investors always wanted low wages and high prices but Saul unlike a lot of founders he was completely uncompromising he was just one of those people that would not corrupt you couldn’t talk him out of it it defended his ethos at every turn drove investors crazy so to avoid all these problems he decides to take the company private which he does the new investors unfortunately are just the same they are captured by the same financial gravity so as a public company he had the same problem as a private company has the same problem one day in 1975 after Saul has been building FedMart for more than 20 years into this huge retail success story he comes into work one day and the locks on his door have been changed he doesn’t work there anymore

Eric Reis (00:17:35):
The board fired him because they were like, this guy is in the way of our faster profits. They wanted faster growth. They wanted retail best practices, not this distinctive, bizarre thing that he was doing. So what happened to FedMart? Will you be surprised to learn that within seven years it was bankrupt? Of course not. Because by betraying the promises that it had made to its customers in pursuit of faster growth in the name of profit, these investors literally enacted the parable of killing the goose that laid the golden egg. But what happened to Saul Price? Well, like a classic entrepreneur, like many of the people we know in common, he took two weeks off and then he was back at work. He actually leased the office upstairs from FedMart headquarters and created a new company he called The Price Club. When I was growing up in San Diego, the Price Club was a local fixture. That’s where we shopped for all kinds of stuff. And FedMart was already a distant memory. However, most of your listeners will never have heard of the company Price Club. And that’s because one of the people that left FedMart to go to Price Club with Sol was a guy

Eric Reis (00:18:28):
Dr. Justin Marchegiani You may remember him from our early discussion about Costco. When Jim Senegal left the Price Club to form a new company, he called it Costco. And actually the company that people today refer to as Costco is actually the result of the corporate merger between the original Costco and Price Club. So Saul Price’s ethos still powers that $400 billion giant to this day. But what Jim added, if you think about like the book Incorruptible is basically the story of Saul Price and Jim Senegal in one package. The ethos of Saul Price, the uncompromising commitment to be a fiduciary to the customer, to build a trustworthy organization. But what Jim added was what I call a governance fortress. He understood that because this corruption is so prevalent, because financial gravity is so powerful, he needed to build an organization that was strong enough to resist I call this organization with structural integrity Costco can make and keep its promises because if you come at the king you best not miss they have the tools they need to

Brian Bell (00:19:26):
resist incredible yeah I didn’t I did not know that story so I mean if this pattern keeps repeating like why hasn’t the market figured it out like why haven’t we figured it out as private equity leader VCs and uh and investors and just we just haven’t been able to figure this out this is one

Eric Reis (00:19:43):
of the most important questions you can ask because we assume that if it was true that mission driven purpose driven companies outperform in the marketplace then by Darwinian natural selection that’s what we would see the economy would be populated with those companies and the extracted ones would lose yet that’s not what we see So we all conclude, therefore, the extractive way of business must have an advantage, a competitive advantage. But this is wrong. We’ve made an error. We’ve assumed without evidence that the market rewards value creation. That’s the Darwinian pressure. And although you could build a market that rewards value creation, we know how to do it. Today’s market absolutely does not. And so Founders are increasingly naive about this point. In every generation, I show 200 years, literally I have a chart of 200 years of these examples in the company going back to the 19th century of founders who thought that by discovering this more enlightened way of capitalism, the market would reward and spread as if they’d invented a new technology. And yet every single time Investors outside pressure finds a way to dismantle these companies at the very moment of their success because this is the naivete. Success makes you a target. It doesn’t just give you freedom and power. It makes you worth capturing.

Brian Bell (00:21:05):
so founders who and then you also succumb to financial engineering well yeah of

Eric Reis (00:21:09):
course because it’s just it’s just so convenient it’s a way to make a quick buck care about the consequences you can make a quick buck this is done in the name but our modern ideology is called shareholder primacy the theory is that this system of maximum

Eric Reis (00:21:21):
Extraction Benefits Shareholders but it honestly doesn’t and we have the data for it now like we’ve been at this long enough that we can see the evidence that this is bad even for shareholders that when shareholders get locked into kind of a it’s almost like a prisoner’s dilemma where since they think every other investor

Eric Reis (00:21:35):
Dr. Justin Marchegiani So we don’t have like we can break out of this trap.

Eric Reis (00:21:38):
It requires First, before investors change their behavior, first founders have to change their behavior. Leaders have to change their behavior and reject the governance best practices that we’ve all been taught. One of my favorite factoids in the book, there’s a study that shows this is now not an insignificant sample. Since 2008, among all public companies, companies that are rated to have bad governance have outperformed companies rated to have good governance. So what are we doing? Where do these best practices come from? Why are people so enamored of them when they consistently destroy value?

Brian Bell (00:22:18):
And is it just the quarterly structure where the finance people get in and they’re like, hey, if I can just juice earnings this way in this kind of PE ratio, then this next quarter I can get a nice little 10-20% bump in the price of the stock and then I can make a really quick buck, I buy some options and then I can even juice the returns even more. So it just turns into a financial engineering exercise instead of a value creation exercise.

Eric Reis (00:22:43):
What’s sad about it, if you look at the data for public companies, if the CFO or the CEO have stock options vesting in a given quarter, R&D spending will be lower. Think about how big the data set of all the public companies, think about how big the magnitude of that effect must be to show up in an

Eric Reis (00:22:59):
Dr. Justin Marchegiani And I think that’s a good question.

Eric Reis (00:23:13):
short term in its orientation. Quarterly guidance destroys value. Independent boards destroy value. The whole notion that the ESG movement has been selling, that we should score companies on three dimensions, environmental, societal, and governance, is absurd because the governance scores are negatively correlated with the environmental scores because today good governance means shareholder primacy and shareholder primacy is incompatible with the kind of long-term investment that’s needed to build sustainable technologies. So I think we’ve kind of made a civilization scale. I would even say we are past shareholder primacy.

Eric Reis (00:23:53):
We’re now an extraction primacy. And it’s time to get rid of that. I would recommend we make a quick pivot to what I call mission primacy and start recognizing that companies that are built to do a specific thing are more valuable in addition to having all kinds of better positive externalities.

Brian Bell (00:24:12):
So one of the most interesting ideas in the book is that organizations become super organisms with their own will.

Eric Reis (00:24:18):
Can you explain what that means?

Eric Reis (00:24:19):
Sure. This sounds very supernatural to people who are hearing it for the first time, but actually organizations are humanity’s first artifact.

Eric Reis (00:24:42):
Dr. Justin Marchegiani And if you’re encountering it for the first time, you should look up something called the Piano Movers Puzzle. Of course, I have links in the book. Piano Movers Puzzle is my favorite demonstrations of emergent intelligence. And here’s what it did. The researchers set up an object that’s kind of like an eye beam. Long side and a short side Connected by a middle bar It’s a rigid object And the task is to move it In between a narrow space Like when you’re moving a couch Through a stairwell Requires a very precise set of movements To get this thing through the gap It’s quite difficult Hence the piano movers puzzle That’s what it’s called

Eric Reis (00:25:13):
So they gave this task to ants. Now, a single ant cannot solve this problem, but a single human can. But what about groups of ants? Well, it turns out the more ants you add to the puzzle, the faster the ants solve the puzzle. And if you watch the video, you will see the ant swarm acting like a person. You can watch it try stuff. It looks just like the person solving the puzzle. That is because the ant swarm, the ant colony has emergent intelligence. It has an intelligence that does not exist in any ant. That’s a well-known finding that this puzzle didn’t prove it. This is just a great demonstration and it makes a really cool video to watch. The more important finding for our purposes is that groups of humans also exhibit this emergent property. A group of human beings is an emergent intelligence in itself and scientists have been able to show

Eric Reis (00:26:02):
That the resulting object, the organization, has a character, it has an intelligence, it has skills and weaknesses that are not present in any individual human. In one of my favorite studies, researchers were able to show that companies have an ethical character that can be measured. The ethical character of the company predicts future compliance violations better than knowing anything about the individual ethics of the individual employees that work there. So it’s a collective phenomenon. I personally think this means that organizations are literally alive. That’s how every founder I know thinks about their organization. We don’t own companies. We birth them. Entrepreneurship is closer to motherhood than it is to slavery. These things are not our slaves that we command. They’re beings that we nurture. And just like with any living creature, the health of the thing cannot be commanded it can only be cultivated so we as leaders have to start seeing our job in building one of these beautiful things as nurturing it to have healthy strong you know strong bones and the inner desire to do the right thing as we

Brian Bell (00:27:06):
define it it’s fascinating so you know if the company becomes its own organism then who’s actually in control is it what we call culture company culture

Eric Reis (00:27:15):
Yeah, I tried really hard to avoid the word culture in this book because just people that’s so vague. So I use the word ethos instead. Ethos is an ancient Greek word that Aristotle used to mean the character of a person. Well, an organization has an ethos. So go back to FedMark. Saul Price had an ethos. But he was able to instill that ethos into the organization such that employees knew what the right thing to do was even when he wasn’t there. And actually the reason it took seven years and not seven months for the investors to destroy it is because the employees fought them trying to preserve the ethos as best they could. I was going to say something else. To understand why this happens and therefore how to control it, we need to borrow one more idea from the past. There’s a thinker named Mary Parker Follett. who was completely erased from management history. She was a contemporary of Frederick Winslow Taylor. And although he went on to become super famous, she was erased from history. You can take your best guess as to why that happened to him and not to her. But in any event, if you read her writing now, Rediscovered in the 90s. Her writing is so incredibly contemporary. You would think it was written like last year. It’s amazing. She wrote about power with rather than power over. She had something she called the law of the situation. She said, it’s not that the superior commands and the subordinate obeys. No, both the superior and the subordinate together obey the law of the situation. They discover together what the situation requires. She said, the responsibility of a leader is to make more leaders. Come on, you’re telling me if you read that on LinkedIn today, you wouldn’t think that was some business leader talking writing right now.

Brian Bell (00:28:44):
Yeah, 100%.

Eric Reis (00:28:45):
But her most important idea is what she called the invisible leader. She would confound her audiences by saying stuff like this that they thought like Yoda level, like what is she talking about? She would say things like this, Mr. Roundtree, the owner and CEO of the Roundtree Chocolate Factory is not the leader of the Roundtree Chocolate Factory. People would be like, lady, what are you talking about? I mean, his family has owned this factory for a long time. His name is right there on the door. He’s not the leader. Who is? She said, no. He is a very good leader for that factory, but that’s because he’s exceptionally skilled at instilling in his employees a sense of common purpose and the sense of common purpose rather than Mr. Roundtree himself is their invisible leader. And if there’s one takeaway I

Eric Reis (00:29:28):
Ignite Insights Ignite Insights

Eric Reis (00:29:43):
is so powerful it has to be powerful to override people’s actual personal instincts so when you realize that the people are not following you they’re following their sense of common purpose we can then ask the much more important question which is of course how do we cultivate a strong culture company a company where the purpose is aligned with what we want it to be and how can we make that durable so it sticks over time rather than being tied to our personal leadership as long as we’re there

Brian Bell (00:30:08):
So why do founders consistently lose this ethos, transference, this, even if they’re like still technically running the company, is it just because there’s just so much pressure from the financial engineers and they just eventually succumb or like why do they lose control?

Eric Reis (00:30:23):
There’s two mistakes that you see come by. I tell a bunch of stories in the book about founders whose company is out of control. They literally are not able to command the thing that they built. And in fact, I know a lot of founders who’ve left. They leave their company because they’re like, I give up. This is not the kind of place I want to work anymore. And in fact, if you read the stories about founder mode, like read all the interviews that people have done about founder mode, they’re all like, the stories are all structured like this, like the company has all these middle managers and all these processes and it’s just like, and so the founder has to go in there and fire everybody and be like, no, we’re going to get back to the blah, blah, blah, blah, blah. And you never see asked in those stories, well, who hired all those people in the first place? who established all those bad procedures, right? The founder is creating conditions that he himself comes to loathe and has to seize control back. So you can see even in that story, there’s an understanding that the company does not automatically reflect your value. So the mistake is first, a lot of people assume that their ethos will become the company’s ethos. So if I’m a bold innovator, the company will be a bold innovator. not necessarily if you don’t cultivate it carefully the company can become very conservative and afraid even if you’re very bold and then you wind up in these situations I tell one of these stories in the book I’ve seen it hundreds of times if you ask people to tell you the story of entrepreneurship like

Eric Reis (00:31:36):
Dr. Justin Marchegiani

Eric Reis (00:31:51):
We need to make a product. To make a product, we must have resources. To have resources, we must have a business model. That business model, we must have a strategy. For that strategy, we must have people. For people, we must build a culture. Anyway, and to then, we build this thing and we offer it to customers. The market validates. our intuition we have product market fit and that’s the role that customers play in our society aren’t you glad that they exist yeah but there’s actually an opposite story you hear sometimes it goes like this the market has a need for a dry cleaner or a florist or a social network or whatever the market has a need the need summons customers to want to buy that requires a product to be built for a product to be built we need employees when we have employees we must have coordination we need managers to have managers we must have a strategy and a business model and a da da da da da da

Eric Reis (00:32:36):
So somebody in the end has to be the kind of the person who got to take credit for this whole phenomenon. Somebody has to be the person who’s like the final decision maker. So this person, the person who privatizes the work of the market. who privatizes the social gains is called the entrepreneur. In this story, entrepreneurs are basically villains who capture social value for themselves. If you ever hear someone say abolish billionaires, you’re someone who attacks entrepreneurship as a vocation, they are in the grip of this story. The weird part about the story is that if you look at the levels of the story, they’re exactly the same, but inverted. So one begins with the founder and ends with customers and the other begins with customers and ends with the founder. It’s the same story told backwards. And in fact many founders, their own executive team is a story two team. They say they admire the founder, but actually they secretly in their hearts, they think this company was always going to be created. See, they weren’t there when the company wasn’t already a success. they can’t remember the struggles to them the company as a monopoly a natural monopoly over the thing that it does so when the founder says hey guys it’s time to pivot we got to go all in on AI or whatever the founder expects them to be like boss you were right when everyone

Eric Reis (00:33:51):
I’ve had the interviews with them when the founder’s not there. They were like, well, maybe he just got lucky. Maybe we shouldn’t go on this crazy quest because we’re going to be giving up

Eric Reis (00:34:06):
Dr. Justin Marchegiani

Eric Reis (00:34:21):
Ignite Insightsights

Eric Reis (00:34:36):
So give us the characteristics of the standard governance. What does that mean?

Eric Reis (00:35:03):
Yeah. So I was like, listen, this company had no protection for the founders. So standard governance means one share, one vote. Whoever can borrow the most money, can buy the most shares, can have the company do whatever they want. There was no mission protection at all. The company was a standard Delaware C Corp. The charter said our purpose is to maximize shareholder value. The company had all these investors who had been longtime partners who were all planning to exit. And they had brought on bankers to run the IPO for them. The bankers had given them these beautiful RFPs full of all these awesome brand name investors who they’re going to bring into the IPO. What they didn’t tell them is that the data shows that the majority, I think it’s 75% of IPO allocated investors are gone within two quarters. so he did this huge roadshow all the investors being like I see into your soul this is the kind of company I want to back all these people were long gone he was really worried after I told him how vulnerable he was but then he talked to his bankers and his lawyers and his VCs and his team his CFO his GC and he came back he’s like you know what they told me they said Eric is such a downer

Eric Reis (00:35:58):
If he really believed in your vision, he wouldn’t talk to you like this. You’re the exception. So he’s like, never mind. I’m not doing anything different. Now, what happened to him? So he could have done something different there. Well, he could have. There’s a lot of things he could have done. Even after the IPO, it wasn’t too late. But the easiest time, it’s always too early until it’s too late. So the earlier, the better.

Brian Bell (00:36:16):
That’s why he’s probably...

Brian Bell (00:36:18):
And before you even raise venture capital 5 or 10 years ago

Eric Reis (00:36:32):
And what’s funny is that, of course, when we see these these dramas, we tend to tell them it’s personal dramas.

Eric Reis (00:36:47):
if you read the stories about this founder they’ll be like he was flawed his business model was flawed it’s like what was so flawed why did you invest first of all second of all even if it’s true that he made mistakes and I’m sure he did did he really deserve so little grace five months really he built this company up from nothing over many years five months at least Saul Price they gave him you know a couple years five years

Eric Reis (00:37:09):
Yeah, but five months. But the worst part about it, you see this over and over and over again. Founder makes a mistake. You see this with Edwin Land at Polaroid. You see it, you know, just like think about when they fired Steve Jobs. The founder makes a mistake. You say, we must have accountability. Fire the founder. Okay, great. But then you don’t actually

Eric Reis (00:37:24):
One in five

Eric Reis (00:37:46):
What are we doing? Everyone thinks they’re the exception, but guess what? You’re much more likely to be in the 80% than the 20%.

Brian Bell (00:37:51):
What did Jeff Bezos do? Because I remember he had letters to shareholders and stuff. He said, hey, we’re going to lose money for a long time, so be okay with that. We’re going to run in the red for a long time, and we’re going to do things that look weird in the short term.

Eric Reis (00:38:06):
He did a lot of things right, but he also got incredibly lucky. First of all, the thing you got to remember is that in the era when Amazon was created companies went public a lot sooner Amazon was

Brian Bell (00:38:15):
public like three years after being founded had like a 400 million market cap I

Eric Reis (00:38:19):
think yeah it was like and raised a relatively small amount of money in its IPO by modern standards yeah so that was a huge thing secondly he was very I think he did well you got to give him credit for he was extremely consistent with his day one letters from the very beginning he really made sure that only investors who understood what he was up to were bought in. He had a very supportive board. He was smart about that kind of stuff. And he just was very clear, if you come at me, we’re not changing, okay? But the most important thing is that he raised a gargantuan amount of money right before the dot-com bubble burst, right before. So yeah, a bunch of runaway. So when everybody else was getting crushed and nobody could raise any money at all, Amazon was flush with cash. And so, of course, he was able to cement his leadership, prove he had the runway to prove that the model worked. And then once it worked, you know, then it was OK. But I wouldn’t say necessarily that Amazon’s our best case study to study because, first of all, it’s very idiosyncratic. No one’s been able to replicate the Jeff Bezos playbook since then. but also Amazon has its own problems because in the pursuit of growth at all costs think about the like incredible human harms that they’ve had in the fulfillment centers and stuff like I think that ultimately is harming the Amazon brand in a really profound way and you just think about what Amazon could be before you know If you’ve read the Cory Doctoral book, you know about inshittification. Amazon’s like one of the big offenders here where if you search for anything on Amazon recently, you get like 20 pages of sponsored results. It’s actually really difficult to get to the first result. And there’s actually research. I forget now, Tim O’Reilly told me about this research, but I forget who did it. There’s research that shows that by Amazon’s own algorithmic rankings, they are pushing you towards the suboptimal product for you. They actually know which product would be best for you, but they don’t show it to you. They show you the worst product. And it’s just like that’s what’s happening at Google. That’s what’s happening to Amazon. It’s happening to a lot of these tech companies where they just can’t resist the allure of the extra money. But again, what does Amazon need the extra money for? They’re just addicted to it for its own sake. They would be creating far more value if they could actually get off that treadmill and restore their leadership principle supposed to be to put customers first. How is it customers first to sell me the product that’s not best for me? Saul Price would be shaking his head if he was still around.

Brian Bell (00:40:25):
Jeff Bezos is probably shaking his head too. So if organizational behavior is emergent and it’s not commanded, it kind of emerges, what does that imply for how we think about leadership? How do we think about leadership?

Eric Reis (00:40:35):
Yeah, so the book is kind of half about the leadership operational challenge and half about what we think of today as governance, the structural challenge. On the leadership side, the most important thing we have to do, and this is going to sound radical to some of your listeners, but it’s actually very practical. We have to do it. We have to commit to a definition of what it means to make a profit. This sounds like everyone’s like, what are you talking about? Everyone knows how to make a profit. Oh, really? I’ve done this exercise with many, many leaders. You say, okay, well, you tell me what it means to make a profit.

Brian Bell (00:41:01):
What?

Eric Reis (00:41:01):
I take revenue minus expenses. That’s profit.

Brian Bell (00:41:03):
Oh, really? Yeah. Okay. I’ll show you my P&L. That’s my E&L.

Eric Reis (00:41:06):
Yeah, exactly. I take a $50 piece of wood. I make a $200 table. I make $150 profit. Oh, good. Excellent. Okay. But let me ask you some questions. Is a Ponzi scheme profitable? Uh-oh. No, why not? Oh, I’m like, we took in $200 in revenue. We only had to pay off $50 to the Ponzi. Is that $150 of profit? No, no, no, no, no. Because eventually the costs will come due, right? Eventually. Meaning if I create value, But I defer the liability. That’s not really profit. I’m just offsetting the revenue minus cost. I’m offsetting part of the formula in time. That doesn’t count. Okay, right. But then what about a toxic waste dump? I’m going to have to clean it up eventually. Is that profitable? People would say, oh, I guess not. I guess not. You work through all those examples, you realize a lot of things we call profitable today really aren’t. Okay. what about what in economics they call negative externality what if I dump pollution in the river and people downstream get sick now imagine I get away with it in all these scenarios imagine I always get away with it is that profitable people if they think about it will be like no it can’t be profitable because although the costs of the dumping are not borne by you they are borne by somebody so just like before we were displacing cost in time now we’re displacing cost in space we haven’t really created net new value so no But what if I created an online marketplace for murder? Murder for hire, okay? Is that profitable? People really want to say no. But if you push, you say, why is it not profitable? They say, well, it’s illegal. I know, but what if I made so much money that I could

Eric Reis (00:42:32):
I’m sorry, I’m sorry

Eric Reis (00:42:48):
but the human life is worth a lot more than $200 so you’ve destroyed one of the input factors of production simple right and just saying like remember I see a kid with a lemonade stand who’s like I made $20 today but they used $80 worth of ingredients to do it right like that’s not profit so that’s just a kid’s table if I take if I steal a $200 piece of wood and I make $100 table out of it I haven’t created $100 of profit I’ve destroyed $100 of value so if our the problem you see in these examples is that we as builders we carry around an intuitive understanding of what I call the builder’s intuition that the best way to make money is to create net new value and then capture some of it for yourself but we also carry around in our head a formalized definition of profit that’s a lot more rigid and over time what we see is these two definitions diverge and then eventually gets us into trouble so the way to solve that problem is just say you know what we as builders we get to decide what profit means for ourselves and so I propose the definition that profit is the maximization of human flourishing nothing more nothing less and anyone who

Eric Reis (00:43:43):
You are a business revolutionary whether you

Eric Reis (00:43:58):
at odds with our dominant finance-driven business culture. So by declaring that we’re going to maximize human flourishing, you not only fix these bugs that infect companies, you also create new opportunities you otherwise would miss. You miss other opportunities. For example, although negative externalities feature prominently in critiques of the profit definition, the bigger loss by far is our inability to see positive externality. things like trustworthiness Saul Price understood that to be trustworthy requires you to follow a principle I call harder is easier in the short term you got to do things that score as ROI negative because the costs of being trustworthy are tangible but the benefits are intangible So we as leaders, if we declare that this is our purpose to maximize human flourishing, we define for people operationally, here is our definition of human flourishing. Here’s how we measure it and reward it. Then here’s our fiduciary commitment. Here’s who we’re going to measure and who we’re going to be loyal to. the people we would rather die than betray and then last we create a condition I call mission drive we make sure that our business model is engineered in such a way that we can only profit by achieving the mission no other form of money making will ever tempt us if you put those pieces together you build an incredibly strong company

Brian Bell (00:45:13):
I love that so you make you make money from the mission fundamentally it’s not just about hey can we can we squeeze an extra dime here in the next quarter it’s hey through our mission and through human flourishing beyond the four corners of our company. I think another way I’ve heard it described is sort of like ecosystem value where you think about a company like Microsoft and one of the most successful companies in history but it created a ton of value outside of Microsoft because you could build on their operating system.

Eric Reis (00:45:46):
So much more value than they ever captured for themselves.

Brian Bell (00:45:48):
that’s right yeah and so you you think about companies like that where they just create so much more value than they’re they’re actually capturing um you could argue whether or not that’s maximizing human flourishing or not but I guess critics would probably say that sounds nice but it breaks capitalism why are they wrong

Eric Reis (00:46:05):
they’re the ones breaking capitalism and I got the I got the receipts to prove it so yeah I won’t do the whole history now because I know we have limited time but This way of thinking about capitalism is so recent. It’s like younger than the trees in your local park. And we have the evidence that shows it is breaking the fundamental premise of capitalism. And the way you know is like this. If you study the history of capitalism, and I mean, go back, read Milton Friedman, read Bastiat read Adam Smith read Joseph Schumpeter read even even John Locke go back all of these people whenever they’re pressed on the question of is capitalism a moral system or not they always like fall back they’re like digging for excuses like well it does this the moral bedrock of capitalism that you hit eventually is this simple idea that when people transact if the transaction is fully informed fully voluntary and uncoerced then both parties are better off when this happens it’s a bit of a magic trick you literally create new wealth that didn’t exist before because both both parties are wealthier than they were before what’s incredible about that

Eric Reis (00:47:09):
What’s astonishing about that is that that wealth was not stolen. It was generated. But notice the preconditions. It has to be fully voluntary, but an addict cannot consent. It has to be fully informed, but if I deceive you, then you weren’t. It has to be uncoerced and how many economic transactions today are from rent seekers or through monopolies or through every kind of coercion you can imagine. So I say that when builders have this intuition about creating value, they aren’t rejecting capitalism. They are the ones defending it. and the people who are trying to financialize everything and make everything into this productive, extractive system, they’re the ones who are causing capitalism’s collapse. Interestingly, Joseph Schumpeter predicted that this would happen in the 1930s. He said one day capitalism will turn

Eric Reis (00:47:57):
Ignite Insights

Eric Reis (00:48:20):
how he could foresee that in the 1930s I have no idea but we’re now living in his

Brian Bell (00:48:24):
prophecy incredible so I know we’re short on time but I mean I could talk to you another two hours on this topic I mean really what are some final takeaways I mean a lot of founders and VCs listen to this podcast they’re growing their companies you know how should they you know what’s the blueprint for them

Eric Reis (00:48:40):
Blueprint is really simple. Just remember path of ethos, path of integrity. Saul Price and Jim Senegal, build something worth protecting and figure out how to build it with structural integrity. You know, we talk about financial gravity. Now imagine a bridge collapses. and I ask you, my engineer friend, why did that bridge collapse? If you say gravity, I’m gonna smack you. It’s like, come on. Yeah, I mean, yes, technically the reason it fell down is because of gravity, but that’s no answer because first of all, it didn’t fall down yesterday, but there’s still gravity, but also there’s 50 other bridges in the world didn’t fall down why this one to study why something collapsed we have to understand the fundamental forces that act upon it we understand load and wind load we understand shearing and tension we check the bolts we’re like oh look all the metal bolts have been corroded and they fell apart that’s why it collapsed because from that understanding we can then say how do you build a bridge that won’t collapse better use stainless steel this time my number one piece of advice to founders is go looking for the equivalent of stainless steel what are the corporate practices what are the the governance structures and the operational disciplines that have that incorruptible character to them and don’t take my word for it about the book is loaded with research but like ask around look around don’t just take for granted that the people you’re talking to the best practices they’re offering you are actually any good for a lot of them we have evidence that shows they’re quite terrible

Brian Bell (00:50:01):
So we’re living through a huge platform shift with AI, right? And how do you think AI either helps this problem or makes it worse or is benign?

Eric Reis (00:50:11):
Both. AI is an amplifier. So just like I said, you know, corporations are are the first AIs on this planet. So they have the same architecture as the transformer. They’re fundamentally emergent intelligences, but the generative version is way faster and way more impactful. So whatever problems you had before, you’re going to have the much, much, much, much worse with AI. But on the other hand, whatever strengths you have can be much, much, much better. In the book, I talk about the structure of Anthropic and the story of how that company came to be. And part of why Anthropic has had the strength to resist when other companies have fallen already into temptation is because I think in part because of its structure. So yes, I think there’s a lot to figure out with AI, but I think fundamentally AI is a governance problem. In AI they call the alignment problem, which is like how do you get the AI to align to human values has a deeper problem behind it, which is how do you align the aligners? We want AI to be aligned to human values, but which human values? If the company making the AI is extractive by nature, the software it makes will be extracted by nature too. We’re already starting to see that in these products. But if you look at the polling on AI, customers and enterprises both Desperately want trustworthy vendors. They want to buy from someone they trust because your relationship with an AI vendor, it’s like a life-saving medicine. Think of the power. If I have a life-saving medicine that you need from me, I have tremendous power over you. If you have all my data, if you are in all my workflows, I have a dependency on you that is really equivalent. And so I think the AI companies that are going to win in the long run are the ones that can prove that they are the trustworthy stewards of this technology. And if the others win, they’re going to provoke a backlash like you can’t imagine that will provoke, I think, a really dangerous and very negative regulatory regime for AI. So yeah, hopefully the good guys win here.

Brian Bell (00:51:55):
So where can folks find more about the book online? I mean, obviously, you’re going to go on the book tour and it’s going to be any place you can buy books, but where can folks find you online and find this book online?

Eric Reis (00:52:05):
Oh yeah please go to incorruptible.co we have links to all the stores that are carrying it which is pretty much all of them plus lots of independent bookstores all across this country who are carrying the book indie bookstores are so valuable as a community resource and they’re under such pressure like if you really want to make an indie bookstores day maybe walk

Eric Reis (00:52:20):
We have readers guides and more importantly for founders, we have implementation guides you might find useful.

Eric Reis (00:52:44):
for many of the techniques in the book. So yeah, do come to the website. We’ve tried to give you every possible incentive to pre-order and to sign up for the mailing list. Much appreciated to those who support the book.

Brian Bell (00:52:53):
Well, I really enjoyed it. The book is incorruptible. Thanks for coming back on. I hope it’s a huge success.

Eric Reis (00:52:58):
Thank you so much. Really appreciate your help.

Subscribe now

Attila’s path into marketing data started with a teenage side project. As a young cyclist, he built a basic webshop for his father’s business. When the first sale came in, his first reaction was not celebration. It was curiosity: why only one? That question pushed him into analytics, consumer behavior, conversion, and eventually digital due diligence.

Today, Attila helps companies, investors, and acquirers understand the hidden risks inside digital businesses—from marketing inefficiency to messy data systems to cloud infrastructure mistakes. In one M&A audit, his team uncovered roughly €2.5 million in risk inside an €80 million deal. The lesson: digital risk is not theoretical. It can directly change valuation, negotiation, and outcomes.

The Paid Marketing Treadmill

One of Attila’s sharpest arguments is that many companies are trapped in a paid marketing system they don’t fully understand.

The trap starts simply. A company spends money on Google, Meta, TikTok, or another paid channel. The campaign works. Revenue grows. So the company spends more.

But then competitors enter the same auction. Costs rise. The company has to spend more just to achieve the same results. Over time, the business becomes dependent on platforms where it does not control the rules.

Attila compares this to a bakery bidding on search terms like “fresh sourdough bread.” If one advertiser pays $0.50 per click, another may bid $0.51. Then a larger competitor enters and pushes the price to $1. Smaller players either raise their spend or disappear from that digital market.

That is the treadmill: you keep running, but the economics do not necessarily improve.

For startups, this matters because early CAC can be misleading. A company may look efficient in its first niche or first geography, but that does not mean the same economics will hold when it expands. Attila has seen startups celebrate strong CAC, only to discover that the next market—such as moving from the UK to the US—is dramatically more expensive.

First-Party Data Is the Escape Hatch

Attila does not argue that companies should abandon paid marketing. That is unrealistic. His point is that companies need more control, and the best source of control is often their own first-party data.

Most companies already have useful data sitting inside their business. The problem is that they do not use it well.

His tire example makes the point clearly. If a customer buys summer tires today, that customer probably does not need to see tire ads from the same company for the next several weeks—or maybe even years. Yet many companies keep retargeting people who already purchased, wasting budget and creating a bad customer experience.

The same problem shows up in banking. Attila described receiving loan offers from a bank despite never taking personal loans and consistently using investment products instead. The bank likely had enough data to understand his behavior, but its marketing system was still blasting irrelevant campaigns.

This is not just a marketing mistake. It is an operating problem. Data often sits in silos. It is messy, incomplete, duplicated, or missing key context like dates and behavioral signals. Without clean, centralized, usable data, personalization becomes impossible.

The Cloud Credit Trap

Attila also warns founders about another hidden startup risk: free software and cloud credits.

Startup programs from major cloud providers and software companies can be helpful. Free credits make it easier to launch, test, and scale early. But they can also create bad habits.

Founders may build on infrastructure that is oversized, poorly configured, or unnecessarily expensive because the bill is hidden by credits. When the credits expire, the company suddenly faces costs it never designed around.

This is especially dangerous because early technical decisions often compound. A stack that looks “free” at seed stage can become expensive technical debt by Series A or Series B.

The question founders should ask is not just: Can we get this tool for free?

It is: Will this still make sense when we are paying real money for it?

What VCs Miss in Digital Due Diligence

For investors, Attila argues that digital strategy deserves more scrutiny.

Traditional diligence often looks at market size, revenue growth, customer concentration, product differentiation, and team quality. Those matter. But Attila believes investors often miss the market’s digital footprint.

That means understanding how customers actually search, compare, discuss, and signal demand online. Search behavior, sentiment, category growth, geography-specific interest, and platform dynamics can all reveal whether a startup is riding a real market wave or merely selling into a narrow pocket of temporary demand.

This is especially important for timing. A startup can have a strong product and team, but if the market is not ready, growth will be harder and more expensive. Conversely, a startup entering a market with rising digital demand can ride a tailwind others have not yet noticed.

Brand Is a Resilience Mechanism

Attila also pushes back on the shallow definition of brand.

Brand is not just a logo, color palette, or tagline. Those things matter, but they are not the core. To Attila, brand is about connection. Real connection with an audience creates resilience.

His example: Apple could make unpopular product decisions—like removing ports from MacBooks—and still survive because the brand had deep customer trust. A no-name company making the same mistake might not survive.

For startups, this matters because performance marketing alone is fragile. If customers only know you through paid ads, you are vulnerable to rising CAC, copycat competitors, and platform shifts. But if your audience has a real relationship with the brand, you have more room to recover, adapt, and compound.

In VC terms, this connects to category creation. The best startups do not just sell into a category. They define one.

AI Will Make Marketing Worse Before It Makes It Better

One of Attila’s more provocative points is that AI may initially make marketing worse.

Why? Because many companies are using tools like ChatGPT and Claude lazily. They ask generic prompts, accept generic outputs, and publish campaigns that sound like everyone else’s campaigns.

The result is sameness.

As more companies rely on default AI-generated messaging, differentiation may collapse. Ads, emails, landing pages, and brand copy will start to converge. Customers will see more noise, not more relevance.

Attila’s view is not anti-AI. The better path is to combine AI with proprietary customer behavior data, market signals, and sharper human judgment. AI can accelerate iteration, personalization, and campaign testing—but only if companies feed it something more distinctive than a generic prompt.

The Investor Question Founders Should Be Ready For

Near the end of the conversation, Attila offered a question he thinks more investors should ask founders:

If a similar company appears in six months, how will you react?

It is a deceptively strong question.

It tests more than competitive awareness. It reveals whether the founder understands their moat, distribution edge, data advantage, brand position, and speed of execution. A weak founder answers with vague confidence. A strong founder has a specific response.

For startups, this is the real challenge. It is not enough to grow while the market is quiet. You need to know what happens when competitors notice the same opportunity.

The Bottom Line

Attila’s message is blunt: growth is not just about spending more, moving faster, or trusting platform dashboards.

Startups need to know where their data lives. Investors need to understand whether CAC is sustainable. Founders need to think beyond the first beachhead market. And everyone needs to be more skeptical of digital strategies that look good only because no one has audited the underlying system.

The companies that win will not be the ones that blindly pour money into paid channels. They will be the ones that understand their data, own their audience, define their category, and build growth systems that can survive competition.

👂🎧 Watch, listen, and follow on your favorite platform: https://tr.ee/S2ayrbx_fL

🙏 Join the conversation on your favorite social network: https://linktr.ee/theignitepodcast

Chapters:
00:01 — Intro to Attila Tóth and Cognitive Creators

00:25 — Attila’s origin story

00:29 — From teenage cyclist to accidental web builder

02:54 — The first online sale

03:10 — Discovering analytics, tracking, and consumer behavior

04:29 — Launching Sight Doctor at 18

05:00 — Early startup failure and hard lessons

05:40 — Digital business modeling for traditional industries

06:45 — The M&A audit that exposed €2.5M in risk

08:57 — Writing Hyper and the frustration behind marketing data

11:14 — The rising cost-per-click problem

12:18 — The bakery ad-spend analogy

14:52 — The paid marketing trap

16:43 — The marketing spend treadmill

18:10 — Searching for an escape from platform dependency

20:00 — Turning years of experiments into a book

22:04 — Self-publishing Hyper

22:34 — Defining the marketing data trap

24:00 — First-party data as the escape plan

24:22 — The tire purchase example

26:29 — Banks, bad segmentation, and irrelevant offers

28:26 — Data silos inside large companies

31:00 — B2B marketing stacks and startup tooling

31:40 — Why there is no perfect tool list

32:35 — The hidden cost of startup cloud credits

34:04 — Questioning the tech stack after credits expire

35:33 — What founders and VCs misread in campaign performance

36:08 — CAC sustainability beyond the first beachhead

38:35 — The UK-to-US expansion problem

40:16 — Digital brand value as startup resilience

42:29 — Brand connection beyond logos and colors

44:01 — Category-defining startups

44:43 — Slack, Teams, and category creation

46:43 — The unsolved interoperability gap

47:50 — Market digital footprint as a VC diligence lens

50:49 — AI, marketing sameness, and lazy prompting

53:41 — The coming wave of AI-generated marketing noise

55:12 — Iteration, personalization, and AI-assisted campaign testing

56:28 — Event-driven marketing and localized campaign signals

Transcript

Brian Bell (00:00:53):

Hey everyone, welcome back to the Ignite Podcast. Today we’re thrilled to have Attila Tóth on the mic. He is the chief strategist and co-founder of cognitive creators and the author of Hyper the untold story of marketing data a book that reframes how marketers, founders, and investors should think about measurement data and digital strategy in an era of rising automation and privacy constraints thanks for coming on Attila sure happy to be here. I’d love to start with your origin story what’s your background

Attila Tóth (00:01:19):

Well, it’s a story that’s not connected to the world of digital. It’s a story that’s related to cycling. I used to be a professional cyclist when I was a teenager and my father, basically my My family were my sole sponsor. My parents were paying for the bikes, for the trips, for the equipment, whatever you need for the races. And my father is an entrepreneur and back then he started a new business. I was in high school. school 11th grade focusing on IT and English and my father thought that my son is learning IT let me give him a summer assignment and the summer assignment was I need a webshop it’s like that we are learning Turbo Pascal and see it’s nothing to do with with web development or anything like that but he was like I really needed for this new business and I said okay I had this internal motivation that they helped me go enjoy my biking and go to different races in country and out of country so that’s the least Amount of thing I can do to at least try it and if I’m not able to do it okay I’m going to say I failed but I want to give it a try so the summer went by and of course with the help of friends Google, YouTube, you name it. I made the webshop happen. Nothing fancy, but it was functional. It had a checkout process and basically gave the keys to my father, the login to the admin panel and forget about the project. And I still remember the day it was raining quite heavily and my father came to my room and said, we made the first sale. First sale? Where? What? It’s like, well, on the website you built, the website I built, that scrappy little thing, I barely made it happen. Yeah, we have the first sale. So then I asked, okay, why just one? how many people visited the website soon I realized there’s no way to track on that website the visitors or what they do on the website so I started learning about data tracking data analytics then asked my parents to buy a couple of books about consumer behavior the psychology of how people think around making decisions making buying decisions so next to cycling quite soon as a hot-headed teenager I had a new hobby that was connected to data and digital behavior. So this is how my story started at 17. At 18, I launched my first business. It was called Site Doctor. The idea was that I can heal websites to convert better. It’s a stupid teenager idea. And yeah, it was an interesting try. A couple of co-founders who were my best high school friends and failed massively after roughly two and a half years. We went basically bankrupt. We needed to stop the business but yeah it was an interesting learning and one of the first failures then which put me on the path of learning more deeply digital business modeling which is a fancy one A way of saying, helping old fashioned businesses create new revenue streams with digital products, digital services across automotive, tourism, healthcare. So really, I would say the boring Big traditional businesses who have nothing to do with digital and then helping them find an edge and build these new services and that’s what I’ve been doing for quite a few years and long story For short, coming to what my focus is today. Roughly a decade ago, we had a client who we built a successful digital business model. They were making a couple of millions of added revenue from this digital services we set up for them. And they called me and said, well, we have now an MP. deal we are buying a scale up and could you check these guys because you built our digital models and this new company is digital first so it would probably make sense for you guys to check it and we did check it We had a service level agreement and said, sure, we’ll use a couple of hours from there. So we check it and we found massive risks worth back then, roughly around two, two and a half million euros at 80 million deal. which basically helped them negotiate better price and it’s really interesting that those risks were not even something that the founders of the scale wanted to hide those were coming from just growing too fast the business and not understanding some basic principles in cloud architecture and technology and creating some data redundancies which were not necessary, which then created massive cloud build service consumption on Microsoft Azure. Again, something we immediately spotted because when you’re a builder, you have to build first with the minimum necessary and then scale it from there. These guys who basically built a really good scale up. It was their second company and didn’t have the experience to spot those issues. But at the end of the day, our client was super happy. We helped them save 2.5 million euros, which even back then was a serious amount of money and this is my focus today that was the light bulb moment that not only building digital business models but actually auditing them on different transactions whether that’s a series a investment or an a or a carve out or a larger merger. It’s something that worked doing and we call this service digital due diligence, which basically looks at the business from a digital perspective. That’s data, technology, marketing, digital brand value, everything related. So that’s where I’m today. That’s my focus. And yeah, in the meantime, I wrote a book about marketing data because it was a topic. It was frustrating many of our clients, why they are spending more and more back then mainly on Google. Of course, ever since the power shifted to social media, the meta ecosystem, TikTok, but even TikTok. And I think that today there’s a big issue on the market like if you go on these platforms, you basically play by their games, and no one is telling you the rules of that game. So if you don’t understand that game, You likely are overspending, whether you’re a corporate or whether you’re just scaling up your startup. So the book is something that started from a frustration. I wanted to understand why this happening. We did a lot of research, a lot of tests and experiments and understood, okay, what’s happening? How can you better work with your own data? As you said, in the era of privacy, GDPR, it’s an interesting topic.

Brian Bell (00:08:50):

Yeah, fascinating. It’s how I kind of got my start in tech as well. You know, about 15 years ago, I started in marketing demand gen.

Attila Tóth (00:08:57):

Okay.

Brian Bell (00:08:58):

Yeah, I ran marketing for a few VC backed startups. So I was probably part of the problem that you’re describing, you know, just throwing lots of cash at trying to grow startups. And I eventually moved to product for a variety of reasons. One was I realized that after running so many multivariate tests, I had to actually go in and change how users got to the aha moment inside of a product, which was kind of what you might call like growth product management today. where that term did not exist. And we might have called it growth hacking back then, like 2013. Let’s just dive into the book. It’s hyper and the current state of marketing data. What year was that book written? Who was the intended reader?

Attila Tóth (00:09:37):

So let me share a bit of of the backstory of the book was 2018. So this is pre COVID times. And we had a client who was doing really well, the digital business model we set up was generating revenue nicely. The only issue we had, and that was a major issue was cost per click marketing spend. So in order to create more revenue each year we have to ask for more money and a conversion rate was not improving meaning if we spend last year 500,000 next year we might need to spend and 550 just to reach the same level of revenue and the reason behind that is because even in 2018 more and more businesses started spending money on the platforms and if we just stay with to simplify the example one platform let’s pick Google search and let’s say you and me have a bakery right in the same town and we advertise freshly baked I don’t know sourdough bread New York okay if you’re the first advertiser then you basically set the benchmark of how much you will spend for each impression and each click I’m the second advertiser and I will try to outbid you so let’s say you spend 50 cents on a click I will outbid you with 51 cents or 52 cents We are two small players, but let’s say a big corporate bakery comes into New York. This is a cheap market. There are only two guys. Let’s just increase ad spend from 50 cents to one dollar because we have the budgets, right? So we, the small guys, we have basically two options. Either we increase our ad spend to pay from 50 cents or 51 cents to a dollar or maybe a dollar and 10 cents. or we are left out from that digital market. And of course, small players sometimes are left out from the market, but the market is huge and there are a lot of corporates who are just pushing up and up the prices. And this is working kind of like the stock market. So the more players you have for a certain audience and a certain set of keywords, if we stick with Google, then those costs are just skyrocketing.

Brian Bell (00:12:09):

it’s actually a business theory actually you know they teach in school that you want to increase your you want to set your marginal revenue to marginal cost right yeah the marketers will come in and they’ll basically just turn up that marketing spend in every channel until every dollar that they get in contribution margin equals the ad spend exactly and so like marketing ruins everything eventually this is the trap right because you come in and you’re just like like everybody else is spending a dollar so I got to spend a dollar

Attila Tóth (00:12:35):

Exactly, exactly. And I took Google as an example, but it’s the same whether we go for Instagram, TikTok, it’s the same engine. The only difference is that new channels are more for giving. So TikTok used to be cheaper until they have reached a certain level of audience and a certain level of paid clients. But if you just think a big back Let’s say 2007, 2008, Facebook used to be like super simple, almost no ads. I don’t exactly remember which year was exactly when they started introducing ads, but even those channels were mainly organic. Now, if you post something on Instagram or Facebook, it’s basically lost. Nobody is going to see it until you spend some money on it. So you pay for it.

Brian Bell (00:13:29):

And then an insidious thing can happen on those platforms now too where once you turn on the paid spend, you actually hurt your organic traffic as well. Something in the algorithm is like, oh, well, This must not be as popular anymore because it’s just not getting as many of impressions, clicks, views, comments, shares. And so the algorithm will now kind of de-weight you in its prediction algorithm because you were paying for that engagement before. And as soon as you turn off paying for engagement, all the engagement that you had before you started paying for engagement, it goes down.

Attila Tóth (00:14:03):

Absolutely. Absolutely. So that was the pain.

Brian Bell (00:14:07):

That’s the treadmill of marketing spend that you get on.

Attila Tóth (00:14:11):

exactly and basically I had a tough question coming from the CMOs one of our clients said why we’re spending more and not increasing accordingly our our income and I had to explain this whole theory how it works but I felt like Google doesn’t pay me so why I’m why I’m explaining this with like kind of like positivity and enthusiasm and telling the client yeah this is how the market works and this is normal and blah blah it was on one end I knew why we are doing that but I didn’t have a better answer to say how to avoid this situation and that was probably a tough conversation but it started I have a question in my head, like, is this really the game we need to play? Is this the only game we can play? And I wanted to find an answer to that question. And I was not sure what that answer will be. was the underlying question that started the book. So the idea started around late 2018. So we started doing a lot of tests, trying different things, as you call it, growth hacking, different ways to still use these platforms, but do also something differently. Most of these failed because we ended up in the same, playing the same game. So whenever we wanted to scale, We were back to square one and needed to increase ad spend. So we started thinking, okay, how can we get outside the system? What’s holding companies back to get outside of the system? And by, I think it was mid-2000... 21 where I had enough evidence how to basically see the trap recognize it see your capabilities and and basically understand what can you do as a company. You cannot be independent from these platforms, probably a myth, and anybody who promises that probably is a snake or a salesman, you cannot be completely independent however there are ways and mechanics that can help you to have a bit of control on what you are doing and that was the moment when I said okay I have enough proof based on our tests and experiences with clients, what are the things that are working and by then I saw this pattern that can be applied from an SME to a big corporate and it’s functional. I was just finishing I think it was late December in 2021 when I decided okay from next year I have to collect everything what we learned since 2018 and try to put it in a format that can be easily digested and can be understood also by people who are not CMOs, simple business decision makers. So the process started and from 2022 until the beginning of 2020, basically I wrote the main copy of the book and then in early 2024 I sent the draft to a couple of people who I think can challenge my My thinking in the right way. These were people who were marketing professors at different universities who were CMOs at different companies or who were data people and wanted to get like a brutal feedback. What am I missing or what am I not covering? Is it consistent enough? Does it make sense? So I collected quite a lot of feedback, incorporated that feedback, and I think it was summer of 2024 when I launched the book on Amazon.

Brian Bell (00:18:19):

Nice. Did you go the self-publishing route or did you hybrid?

Attila Tóth (00:18:22):

Yeah, self-publishing. I didn’t write this book because I want to be an author. It’s not even in my LinkedIn. I know many people put in their bio their authorize. I don’t consider myself an author, I consider myself an experimenter who wanted to share the pain and learnings with people who want to understand this challenge and want to find a way how to get out of this trap.

Brian Bell (00:18:49):

So let’s talk about some of the learnings in the book. So we talked about the trap. Let’s talk about the escape plan in the book. Define what the trap is a little bit and then we’ll go to just how do you get out of it?

Attila Tóth (00:18:59):

Yeah, so probably in a nutshell, the trap is thinking that by going If you’re going on these platforms and spending on digital marketing, you’re doing the right thing. And you need to always spend more and more and just adapt to market conditions. If competitors are spending more, you try to increase your budgets. and if for some reason they are not covering a niche then you’re happy that you have a niche where your cost per clicks are super low and that’s it and that’s how most marketers behave because these platforms I want you to think like that. Now the escape plan. First and foremost, there’s no magic bullet. However, there’s one common thing that we identified in our tests and experiments. And that common thing is that each company sits on some level of marketing data. And I’m going to tell you like a super simple example. Let’s say you’re a company who sells tires, summer tires, winter tires, of course, right? Most of the companies who sell tires across Europe, US don’t utilize their own customer data. Meaning if I sold a tire to Brian yesterday, summer season is starting and you It’s highly likely that in the next four weeks you won’t buy another set of tires it’s highly likely I’m not saying 100% because you might have a second car but highly likely you won’t buy the same Size and type of tire now most companies don’t utilize that data that Brian just made a purchase yesterday so what will happen that Brian for the next two to four weeks you will see tire ads from the same company You just purchase your tires for the next couple of weeks.

Brian Bell (00:20:56):

It’s like, I just bought your thing and you’re showing me this.

Attila Tóth (00:21:00):

Yeah. And this is probably the easiest low hanging food to take your customer data and make sure you exclude from your campaign those people who recently made the purchase. Of course, it depends on the product, the seasonality. But in case of tires, you can take out a person from the campaign probably for quite a long time because maybe in the US, I don’t know, the average tire...

Brian Bell (00:21:26):

You know if they bought summer tires in April and therefore you don’t need to show them summer tires for at least a year or two, right?

Attila Tóth (00:21:34):

Yeah, at least two or three years. So this is like a really simple example. But the principle is that your company is sitting on first party data that’s yours. You didn’t buy that data set from outside. it’s based on how your customers are behaving right now and you should use that data actively and of course depending of the type of business what actively means can be different let me give you another example banks. Even in the modern era of fintech where we have Revolut, we have Wise and all these fintech companies who are revolutionizing how banks work. Many traditional banks are still stuck on not using I will give you a simple example one of the banks I personally use they keep sending me offers with loans and what they didn’t realize I never took a single loan on my property I always invest my money to different saving accounts or the stock market, and they don’t offer me products related to investment. They try to send me a loan, which is in my case, 99%, I’m not going to take a loan from them.

Brian Bell (00:22:54):

Well, maybe to steel man the other side, maybe they’re looking at it and they’re saying, hey, this guy has the ability to take a loan out. He’s a low risk. And he doesn’t have a loan. He should have a loan. Right?

Attila Tóth (00:23:05):

Yeah. But usually what happens, they just have an email database. Right. And the CMO cell, well, this quarter we need to increase our loan rate.

Brian Bell (00:23:16):

We need so many numbers that he locks by the end of the quarter. Go. Yeah. Yeah.

Attila Tóth (00:23:20):

and they just push the same message for everybody and it’s not working it’s not converting and sometimes it’s even annoying like you get a message that’s completely irrelevant to what you have or what you need so the whole book is based on the principle of how you can utilize your own data and activate it in your digital marketing and growth campaigns. And this sounds simple but it’s not so simple to do. And the reason why because many companies especially the large ones sit on humongous amount of data which is uncentralized living in silos nobody knows where data points are nobody cleanses that data. So it’s a mess. If you’re a startup and you understand the principle, you can already build your company in a way like, I want to make sure that I’m utilizing my first party data strategically. But in most cases, sites or larger companies haven’t built their businesses around that principle. So when you say, okay, let’s activate some of your data points you’re having, like We don’t know where that data sits. Oh yeah, it’s in Salesforce somewhere, but we don’t actually understand where the Salesforce database is. It’s somewhere in the cloud. Okay, we do a CSV export, but that’s a one time and then it’s not working. And of course, today with AI agents, You can automate many things, but you understand the issue. Data is in silos. It’s messy. Most of the data is not useful and it’s not useful because it was collected in a way that some really critical parameters are are missing, like the date, let’s say, when there was an event, let’s say it’s an e-commerce platform, and somebody put in their wish list an item, right? And if you don’t have the date when that wish list item was saved, that data is almost useless. Because if that event happened two years ago has a different meaning compared to if it happened two weeks ago. Again, depending on the product data, whether it’s a B2C Yeah,

Brian Bell (00:25:36):

and this is why Marketo was so powerful when it came out, you know, 15 years ago to almost 20 years ago, is you could go in and start creating smart lists and static lists on the customer data. So I remember doing a lot of those CSV exports and merging and list creation and segmentation and what are you seeing in best practices? This is very much a startup podcast, right? Venture Capital and Startup Podcast. What are you seeing out there on the B2B front that’s working like in today’s marketing stack?

Attila Tóth (00:26:08):

I think it’s a stupid answer but it really depends on what type of B2B startup we are talking about because there are so many tools out there and I don’t think there’s a list of perfect tools how I approach things as we do a lot of digital due diligence whether those tools make sense for that specific company in the growth phase they are so if it’s series A it’s slightly different are different from a series B or from a fully grown corporate merger. So they need to have a different tool set. What I see, and this is, I think, thanks to, on some level to COVID, during COVID, Most of the big guys, Microsoft, Adobe, gave a lot of credits to different accelerator programs, venture capital companies, credits that startups can use to just buy a license, We set up basically free of charge a Microsoft Azure infrastructure, which sounds super sexy, like you get $60,000 of credits for a Microsoft environment, right? Sounds really good. But what happens when those credits and usually these guys don’t really have an incentive to help you set up your ecosystem in a smart way they have the incentive to give you those credits because they think okay if I don’t know from a thousand startups only a hundred will grow and become a serious company and maybe one or two becomes a unicorn then that the math is is worth it for them But the rest of the companies who are maybe growing slowly or don’t even have the vision to become a huge corporate will pay more than they should for services they don’t need. And that’s a trap, a different kind of trap. I see that many startups use these credits because they are free of charge, but once they are out of the credits and start billing and they have some revenue, they never question whether the tech stack or the tools they are the right fit for the stage the company is going to. So it’s hard. There’s no magic bullet here as well to say these are the five marketing tools I would use. What I would recommend in instead of thinking how can you capture your own data in a centralized way and if it’s a custom-made agentic AI that grabs from your website forms the data points and puts it into an air table it’s fine it doesn’t have to be necessarily fancy it has to work until you of course reach a limit of growth where then you say okay this was the bootstrapped version now we do the next layer but even the next layer shouldn’t be sales for 360 automation which again could cost a lot of money what’s a common

Brian Bell (00:29:17):

mistake you see founders making when or VCs when they’re doing due diligence on interpreting campaign performance

Attila Tóth (00:29:23):

Looking at the first mistake that’s really common, let’s say this is something actually we just finished an audit on a series A startup and they had a really good customer acquisition cost. It made a lot of sense financially speaking and they were super happy.

Brian Bell (00:29:43):

So your CAC, your customer acquisition cost was way lower than your lifetime value. You’re paying back very quickly. The VCs are happy.

Attila Tóth (00:29:51):

exactly and nobody questioned whether that cost is sustainable and by sustainable I mean for those audiences they are attracting in this quarter are those audiences big enough and will will it cause the It’s the same when you scale for a different niche or a different beachhead market because of course in the startup world first you want to conquer your first beachhead and then you go to the next one. The reason they’re not asking this question is because they are thinking only short term and we have to survive the next 12 months or the next 18 months and then we prepare for the second round of investment. And that mindset, of course, works well when the economy is booming and everybody’s throwing money at VCs and startups and VCs have larger funds. But in the current circumstances, I think there’s a lot less capital deployment on the market, especially for Series A and below Series A, then you shouldn’t think of your first round or your second round as just a breath of fresh air to survive. You should think a bit ahead at least two years three years and to understand the market dynamics because I saw at least so in the last six months I saw five companies who were super happy with their CAC were not looking looking wide enough on the market and one of these companies already basically reached the audience that’s capable of reaching and now they’re asking okay for the next frontier we have to go here and here that’s a different country they started in the UK and now they are attacking the US and the US is way higher in terms of customer acquisition cost and the ROIs are really bad yeah so that’s that’s a common common thing that

Brian Bell (00:31:52):

I don’t know if you agree with this statement but I’ve noticed that CAC doesn’t tend to go down over time. Have you seen it actually go down over time? It tends to go up, right? So that’s an interesting truism I think in demand gen circles. And something I think as you analyze companies that you should be aware of, if the CAC is such that it takes a year to pay back right now and they’re at like seed or series A, it’s not going to get better than that. You know, maybe they’ll be able to cross sell and upsell and get more revenue out, but the CAC is not going down, right? and so you got to really consider that are they do they have this is something we care about a lot now and I think it’s become the bottleneck in venture capital or startups I should say is it’s distribution right because it used to be like oh like I could I could build something that’s amazing and now it’s not so much about that anymore like anybody can build stuff and engineers are 10x more efficient or whatever it is with cloud code and codex and and others so you can build whatever customers need the real real challenge is getting them to care and care fast enough and pay for it

Attila Tóth (00:32:52):

Exactly. And here’s another thing that I saw it working that many also VCs and startups completely miss is the digital brand value. So branding, I think, is one of the most misinterpreted word out there in marketing if you ask a designer what’s branding they will say it’s the logo it’s the colors if you ask a CEO it says it’s the company values so everybody has a definition of branding and on some level they are all true but I think the most important element of branding is connection real connection not fake connection real connection to that audience and I like to use this example with with Apple so a couple of years ago I think Apple messed up with one of the MacBook Pro versions they removed all the different inputs from HDMI SD cards and they were just going to full flat USB-C however since there was a connection to the brand with their audience they survived this strong decision and then I think three four years ago they brought back these You can see these different input options for MacBook Pros. But if a mistake of that level is done by a no name company in the market, that company can go back up because there’s no connection to that brand. What startups and we see sometimes miss is investing in the brand and understanding how you create that connection which is not this color or logo which of course are important but it’s a more meaningful connection And this is something that many startups don’t focus on at all, like, we want to get product market fit. Okay, but what happens when you start with a product Saturated product market fit. What’s going to happen there? Branding, especially digital branding, it’s a tough thing to crack. But if you don’t start early enough, then you will be just another company in the market. best case yes you do an exit to a PE or a big corporate who can put your company strategically in their ecosystem but you have no brand value that’s something that’s difficult to explain to investors and most founders don’t know how to explain that and that’s the only thing I saw from a data perspective where brand value was measurable and visible CAC I’m not saying it it it become like 50% cheaper to get the customers but it survived rough seasons

Brian Bell (00:35:41):

without going into crazy heights yeah we call that in our our framework of evaluating startups we we call that category defining meaning like can this brand can the startup define the category or own the category in which it’s playing where people synonymize them with that That product or service right and then you get the brand halo effect and that’s yeah it’s hard to predict that as a VC when you’re looking at a startup but it’s it’s something that we do ask ourselves can this define the category will this become the apple of of this this you know b2b segment absolutely there’s a simple

Attila Tóth (00:36:18):

question and this is something I I learned from one of my mentors when you want to Understand a category or a brand. Even if you know your audience and you ask a couple of questions, sometimes they don’t know the answer because they think limited in the box of a different category And that’s also why defining a brand in a completely, or even if it’s not completely new, because let’s just pick Slack messaging apps where they are before. for, right? But Slack found a specific category in which they dominated the market. And if you would ask before Slack people in the corporate world, in startup world, what are your pains in communication? They would have probably different answers, but none of those answers would point to a solution what Slack has done. So it’s a tough thing to... Yeah, I don’t think anybody would have said,

Brian Bell (00:37:21):

you know what, I need an instant messaging app. you know with channels and and I need to be able to hashtag and at sign people inside of channels I don’t think people would have would have said that and then you had teams that came along after that and decided oh you know what we’ll make the calendar the channel and the chat you know integrated with like OneDrive and and so like every recurring meeting can have its own like drive or every channel can have its own drive. And I think that was like an innovation that teams had.

Attila Tóth (00:37:49):

One thing nobody actually solved, and I think it’s a good niche since we do a lot of M&A work. and have a lot of different companies to audit. Let’s say the mother company is on Microsoft, the corporate and the scale-up they’re investing in or the startup they’re buying, it’s on AWS or Google Cloud and those systems just don’t talk to each other. Nobody is solved yet in a good way to synchronize your Microsoft Outlook calendar with your Google Amits calendar. And it’s an interesting niche and I don’t know who will solve that, but definitely if you would ask a couple of people today, they couldn’t define the category, but there’s a problem that’s existing.

Brian Bell (00:38:36):

What other things should VC, early stage VCs in particular, like us look at when we’re evaluating a startup’s distribution and paid marketing and acquisition channels?

Attila Tóth (00:38:47):

One of the things that are most of the time hidden on plain sight is looking at, we call it the market’s digital footprint. Let’s say you have a B2B startup that is in the automotive sector. Post VCs will look at the economics from potential market size, country regulation, so on and so forth. highly likely miss is looking at market dynamics I’m going to give you an example autonomous driving it’s a really cool thing and it’s a category that’s that’s growing however there’s a difference in how Audiences are on one end expecting autonomous driving and the other end trusting autonomous driving. So if you go to a German market and look at the data how people search for cabs, Uber, Bolt and different solutions is different compared to how people search for those services in Japan. And those slight nuances open up niches and sometimes those niches are Really easy beachhead markets where a startup can enter and grow really fast. And I think this lens of looking at the market’s digital footprint is missing from most of the VC checklist. not because they don’t have people who understand data it’s considered secondary because they never saw relevant data in action and I think that’s something where VCs need to Be open-minded and look at some investments happening on a global scale where these

Brian Bell (00:40:44):

data points are accessed Next two questions are kind of related, but you’ve been in marketing for a while, you’ve seen a lot of changes What’s changed the most in the last five or ten years in your mind? And what are you looking forward to in the next five or ten years?

Attila Tóth (00:40:58):

Let me share a funny example. This is an example from US. So there are three different companies in the same segment, competitors. I think until 2022, all of these companies had different brand positioning. So if you looked at their marketing, one is really red, the other one is blue and the other one is yellow. But not just in colors, in messaging, positioning, everything. Now, if you look at those three companies, the branding is different but the visuals messaging is almost identical and of course that’s because Claude ChatGPT and basically this safety and I like to I may be a bit aggressive with this word but I think it’s laziness of using AI tools just on their default mode like okay I’m running a campaign for this audience in this country blah blah blah for this product Yeah. And then of course, how LLMs work, they connect different data points and generate an answer. And if you don’t have anything specific, like you don’t really put in effort and using the LLM to really...

Brian Bell (00:42:15):

You just like take whatever, whatever it’s outputting on 5.4 or 5.5 or whatever it is. Yeah.

Attila Tóth (00:42:20):

Yeah.

Brian Bell (00:42:22):

That sounds good.

Attila Tóth (00:42:24):

And that’s... that’s something I I’m seeing that’s happening and I think in the next five years it will get worse meaning even for a SME like even if they didn’t have a CMO or startup didn’t have a CMO they had a friend or they asked an opinion or like we are building this product could you let’s sit down for coffee I invite you for dinner like you know growth hacking get getting ideas for their marketing now what happens

Brian Bell (00:42:56):

Or even going to a conference.

Attila Tóth (00:42:58):

Or even go to a conference. Yeah. Now what happens, they sit in front of their laptop. Best case, they have a paid version on one of these LLMs. And that’s the best case. and there will be many similar campaigns many similar messages coming out which will automatically destroy customer acquisition costs these will go up conversions will go down so I think in terms of marketing things will get worse first and then of course because there are always innovators people realize okay we need to do deeper work and come back stronger we can use the tools but maybe use again coming back to your own first part data use the customer behavior data you have and feed it in and create your custom GPT not just you could iterate

Brian Bell (00:43:51):

so much faster now like we used to have to manually create ads and manually set up campaigns and manually just adjust everything kind of by hand really and you know you had a tool you had a SaaS tool and then you’d you know run your Marketo campaign with different subject lines or whatever different body messages or different headlines or calls to action and you’d run a multivariate test and it was a lot of work to set it all up and run it and get the results and now I would imagine with the app you can like iterate your way to better messaging perhaps have you seen any of that in

Attila Tóth (00:44:21):

the field yes I know in the sense of laziness we see that most people even if they can iterate they won’t iterate because initially it saves them time just to put out something and the market is not yet crowded with AI slop meaning it’s it’s not yet we are not yet in the worst phase of marketing so it’s still working and in some niches even these basic prompts are working but those will stop working and then yeah so in that sense but yeah answering also to to your question in iteration I saw a couple of Good things as well, specifically in sports fashion. So when you advertise, let’s say, footwear, running footwear, now everybody’s talking about Adidas and the new word record. And of course this is a good advertising moment for them which will certainly pass but what I saw some brands do really well is to use data points from events happening let’s say in in Switzerland near Geneva there’s a trail marathon in the Alps right and they use that event to promote a specific running shoe in that area because they know this event is happening I don’t know mid-June it’s attracting these type of people so they’re using that event as a wave they are they are basically surfing on to create content around that and personalize their communication with different iterations before, during and after that event. Not directly, because of course, this would mean you have to go and sign partnerships with all the events, but indirectly, but just by having a photo from the mountain that’s in the event, or there are like really smart ways, creative ways, like You get a data point and you can creatively think about what does that mean for images? What does that mean for messages? What type of people will be running here? And those kinds of personalized elements can be iterated really nicely in campaigns and drive up conversions temporarily. But that’s the beauty, as you said. Now you can run the same mechanism for a next event that’s happening in Italy or in Spain or in the US and you don’t need you just set up the framework use different data points and generate completely different outputs while giving a personalized experience to the audience it’s not having the same standard message in all the countries but using these little insights to customize and iterate. That’s something I saw. It’s working really nicely with one of the footwear companies that I’ve been advising.

Brian Bell (00:47:16):

Yeah, I’m sure marketers will figure out how to make it like minority report where, you know, it knows exactly my blood type and everything about like how I slept last night. And, you know, you know, because I wear, you know, I wear the Fitbit and, you know, I didn’t sleep well last night. So I’m sure they’ll be like, you know, didn’t sleep well last night. You need some Monster Energy drinker.

Attila Tóth (00:47:34):

That’s another topic in the book, in the last chapter, which I think we should be talking about more as people, what happens with our data. So in an ideal world, I think we should own our own data, as you said, your Fitbit data, your location data, everything. And in an ideal world, you should have a vault, a data vault, where you say, okay, I’m going to share my data with I don’t know this private clinic McDonald’s or the companies you are highly likely interested in what you do and you either share it freely because those services are relevant to you or if you want you can sell that data and you then you are aware I’m selling my data to XYZ company

Brian Bell (00:48:24):

I think I started trying to do this. It always feels too early. It’s like, okay, and here’s what we’re going to do. We’re going to go around all these products and services and aggregate data. And I’m like, okay, great. And then what? Well, we’re going to help people use it places. Okay and then what and it’s just there’s no business there I think it’s kind of like I think it’s there for defined categories like sharing my calendar right I need to share my calendar with certain services for certain reasons but like if you start like taking all my data and putting it into something There’s just so many variations of how that data is stored. It’s a tough one.

Attila Tóth (00:49:01):

It’s a tough one. Even in the book, I... Have you ever downloaded your data from a service?

Brian Bell (00:49:07):

Like I downloaded my Fitbit data. I was interested. So I was like, okay, I’m going to download it and I’m going to upload it in the chat. I have this project for all my personal health and wellness stuff. And so I downloaded it. It’s going to be great. I’m going to download it. It’s going to analyze it. And the way that it was constructed, I expected like a big CSV, like a log, like a table, right? That’s what you expect. It was literally just like plain text files and folders. And so I’d have to run some sort of ETL transformation, which I probably could code up with Codex or Cloud Code or something. But it’s just like, that’s what I’m talking about. It’s just like, there’s so much data out there stored in all kinds of, and then I’m wondering, did they just make it hard on purpose? I think so. To obey the law. Yes, you have data portability, but we’re gonna make you download it in such a nefarious way that it’s almost unusable like my sleep data like you should have it it’s like on this date this many hours this much deep sleep blah blah blah no everything was like a little plain text file like every cell on a CSV was a plain text file in a file folder system. I’m like, well, who does that? Why’d they do that? Anyway, I’d like to wrap up just a few wrap up questions for we run out of time. What’s a question you wish investors ask more often of founders when evaluating their digital marketing strategy? In your own practice, what you’ve noticed, what has been a signal or metric that has become significantly more predictive of startup success?

Attila Tóth (00:50:31):

That’s an easier one. If we are talking about business categories that exist, it’s basically digital trends. So what’s the sentiment around it? How many people are looking into that topic? Is it a topic that’s growing in terms of their trend or decreasing? from simple search terms to complex things like now you can analyze LLM sentiment so let’s say what’s the sentiment around a special kidney disease or you have so many interesting insights that you can capture and utilize as an investor but also as a startup that could help you define and faster product market fit can help you define your brand these are things that are out there it’s almost free of charge and people are just like they’re not using it

Brian Bell (00:51:26):

I would describe this as in our framework is timing. Why is it the right time for this business? And part of that timing is the quantification of what you’re describing of market readiness. Like, hey, there’s a trend happening right now where it’s the right team with the right idea at the right time and the right market conditions. And there’s like a tailwind here of like just market interest in this category.

Attila Tóth (00:51:49):

Yes, and I think why timing is probably the best word to describe it is because even in like really traditional assets like hotels and real estate, I’m just going to give you one example. This is already out of NDA so I can share it we had a client interested to invest in some luxury hotels in in East Asia and wanted to pick a location and most of their advisors were coming with different offers like here the land is cheap or or look at this island with the views and so on. So they were like really emotional and not objective parameters. And what we did is we looked at how many people are searching for different destinations. in East Asia and back then this is like more than five years ago we saw a massive interest spike in the area called Lombok and Lombok was basically underdeveloped there was nothing there I wouldn’t say it’s Spectacular but since there were a couple of people influencers who went there either for diving or hiking or whatever reason they started creating a micro trend which turned visible in google searches and with with all the investors well people are wanting to go here there’s no infrastructure there are no hotels yet so you can just ride that trend Yeah, exactly. I think timing is the best way because now if you look at Lombok, it’s already full. It’s not the right time to invest. But yeah, timing is the perfect word.

Brian Bell (00:53:26):

All right. So last question. Back to the first question. What is the question you wish investors asked more often of founders when evaluating their startups and digital strategies?

Attila Tóth (00:53:36):

It’s hard to pick one question, but if

Brian Bell (00:53:39):

Or two or three, yeah.

Attila Tóth (00:53:41):

If I had to pick two or three, one would be how is your digital strategy compared to similar, we are now, many people call it SaaS Apocalypse and I think it’s not but we’ll see but if I would ask a startup how would you compare yourself to a similar software service startup yes you are now an AI company but what’s different in your digital strategy what sets you apart and usually the answer will be they don’t know if they are on They say something, but in reality they won’t know.

Brian Bell (00:54:31):

I don’t ask it exactly like that, but I say something around what’s your distribution wedge? How can you acquire the next 100 customers better than your competitors? Something like that, some variation of that, depending on how much context I have and how much I know. yeah and I just kind of listen to see what they say you know because I’m always interested because I’m a demand gen person right I want to learn right oh you’re like oh I haven’t heard of that tool what is that tell me more you know sometimes they’ll have something really creative they’ll say something really creative I’m like okay this this person’s like a growth hacker and I think that really it’s a really important skill now it’s really important to like absolutely hacks of growth

Attila Tóth (00:55:08):

I think it’s more important than ever because today most marketing again will be based on cloud or chat GPT output and if you don’t have anything else that’s outside of that and you have basically nothing that sets you apart I have one more in my pocket. Yeah, go for it. The second one would be if a similar company appears in six months, how will you react? And that again tells you on one end the mindset, but whether they have enough preparation and enough data in the background to know, okay, even if we are new and fast now, but competition will come and then what?

Brian Bell (00:55:49):

Yeah, I like that question. I’ve never asked it quite like that. I usually ask it like, who are you worried about? Oh, we’re not worried about anybody. But like, okay, let’s imagine that’s the good follow up. Okay. Imagine somebody did come along and they’re doing exactly the same thing.

Attila Tóth (00:56:01):

Yeah. Yeah.

Brian Bell (00:56:02):

How do you react? I love that. I really enjoyed the conversation. As you can tell, I could probably talk about this for another hour or two, but we’re out of time. Where can folks find you online?

Attila Tóth (00:56:10):

LinkedIn. That’s the, I think only social media platform I actively use LinkedIn slash in slash creative Attila. Awesome.

Brian Bell (00:56:19):

Thanks Attila for coming on. I really enjoyed it.

Attila Tóth (00:56:21):

Thank you for having me.