What Happens When a Curious Teenager Accidentally Helps Invent the Cloud?
Imagine a long-haired kid in a tiny Seattle computer store in the late 1970s, opening boxes of manuals he was supposed to shelve—but instead reading cover to cover. Customers wandered in, confused. Employees shrugged. “Ask that kid in the corner.”
Fast-forward a few decades and that same kid—Jeff Barr—has quietly helped explain, shape, and humanize the cloud for millions of developers as AWS’s Chief Evangelist.
The Career Pattern Nobody Plans (But Everyone Lives)
Jeff’s story isn’t a master plan. It’s a string of “that looks interesting” decisions:
Teen job at a computer store → deep curiosity
Hardcore engineering roles → love of building
Consulting in early web services → explaining complex ideas
Accidentally discovering Amazon’s first APIs → click
Becoming AWS’s first true developer evangelist → history begins
Zoom out and you see the pattern: deep understanding + clear explanation = leverage.
Jeff didn’t chase titles. He chased clarity and his curiosity.
And that turns out to be a cheat code.
AWS Didn’t Win on Marketing. It Won on Trust.
One of the most surprising revelations: early AWS barely had marketing.
So Jeff did something radical in 2004—he launched a blog.
Not “thought leadership.” Not “brand voice.”
Just one engineer explaining new services, honestly, in plain English.
Over 20 years:
~3,300 posts
~1.5 million words
~150+ service launches explained by someone who actually used them first
The secret wasn’t hype. It was credibility.
Jeff insisted on:
Hands-on access before writing
Saying what wasn’t ready yet
Killing “enterprise speak” on sight
Writing to developers, not at them
Result? Developers trusted AWS before they trusted the cloud.
“Create Your Own Luck” (A Masterclass in Boldness)
One of the best stories sounds fake—but isn’t.
Jeff once followed Bill Gates into a McDonald’s to ask for early access to an operating system his company needed. It worked.
The lesson wasn’t recklessness. It was preparedness.
You can increase the odds of lucky moments.
But when they arrive, you must recognize them—and act.
Luck favors people who:
Put themselves near interesting things
Know what they’ll ask before the moment comes
Are brave enough to ask once
Swiss cheese holes line up more often than we admit.
The Cloud’s Hidden Impact: 10× Lives, Not 10% Raises
Jeff now spends most of his time traveling—meeting AWS communities across Latin America, Asia, Eastern Europe, and beyond.
The pattern is humbling:
Developers self-organize
Study together
Certify together
Change their families’ economic futures
Not 10% better.
10× better.
Cloud skills + community + persistence = social mobility at global scale.
No deck captures that.
AI Changes the Developer’s Job (But Not the Responsibility)
Jeff has lived through mainframes, PCs, the web, SaaS, cloud—and now AI.
His take is refreshingly sober:
AI coding assistants are just the next tool
Writing code matters less than reading and validating it
The commit still has your name on it
The real new skills:
Explaining intent clearly (to humans and machines)
Reviewing generated code critically
Communicating requirements with precision
The future developer looks suspiciously like a product manager with better instincts.
The Next Era: Disposable Apps, Durable Data
One of Jeff’s most provocative ideas:
Apps may become disposable.
Data will not.
With AI + serverless:
Build fast
Use briefly
Throw away
Rebuild when needed
What lasts?
Clean data
Clear specs
Deep understanding of users
Infrastructure becomes permanent. Apps become fast fashion. Insight becomes the moat.
What Jeff Barr Wants to Be Remembered For
Not scale.
Not AWS’s size.
Not even cloud.
Just this:
“I gave people accurate information that helped them grow.”
That’s the quiet superpower behind every enduring tech platform:
someone who cared enough to explain it clearly.
And once you see that pattern—you start seeing it everywhere.
👂🎧 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 – Welcome to the Ignite Podcast
00:45 – A Teenager, a Computer Store, and the Birth of Curiosity
02:45 – Learning by Reading the Manuals (Literally)
04:10 – From Retro Computers to Real Engineering
05:45 – Startups, Microsoft, and the Pain of Big Companies
07:30 – Web Services Before They Were Cool
09:50 – The Accidental Discovery of Amazon APIs (2002)
11:30 – The First AWS Developer Conference (Before AWS Existed)
13:30 – “I Have to Be Part of This”
15:00 – Joining Amazon at Its Lowest Point
17:00 – Inside Early AWS: Less Structure, More Vision
19:15 – Becoming the First AWS Evangelist
21:00 – Launching AWS Through a Blog (A Radical Idea in 2004)
24:00 – Writing for Developers, Not “Enterprise Speak”
27:30 – 20 Years, 3,300 Posts, and 150+ Service Launches
30:00 – AI Coding Assistants Are Just Another Tool
33:00 – Reading Code Is the New Writing Code
35:45 – re:Invent Takeaways: Community Over Everything
38:30 – How Cloud Skills Change Lives Globally
41:15 – The “One-Person Unicorn” Thesis
43:45 – From Infrastructure to Agentic Applications
46:15 – Disposable Apps, Durable Data
49:00 – The Developer of the Next Decade
51:30 – Jeff Barr’s Legacy
53:15 – Closing Thoughts
Transcript
Brian Bell (00:00:58):
Hey, everyone. Welcome back to the Ignite podcast. Today, we are thrilled to have Jeff Barr on the mic. He’s been helping shape the story of cloud since the beginning before serverless was cool. Currently serving as chief evangelist at AWS. He writes, builds, evangelizes, experiments, and pushes boundaries on the cloud while keeping one foot in the hands on tech and other narratives. Fresh off reInvent 2025. Welcome to the program, Jeff.
Jeff Barr (00:01:21):
I’m really happy to be here, Brian.
Brian Bell (00:01:22):
I’ve been wanting to have you on the pod for a year or two now. I’d love to get your background and your origin story. How’d you end up in cloud?
Jeff Barr (00:01:28):
Wow. Let’s see. So I was super lucky that my actual first job in tech was when I was a teenager. And when I was 16 years old, I worked at a computer store up in Green Lake, not too far from where we are in Seattle right now.
Brian Bell (00:01:41):
Fond memories of Green Lake growing up in Seattle. Yeah.
Jeff Barr (00:01:44):
So I was on Northeast 72nd Street in Green Lake in the late 1970s. And a friend hired me into the store. And my official role, if I even had one, was when all the books and magazines and literature came into the store for the brand new world of personal computers. My official job, I was supposed to just slice open all those boxes and put all that literature and great content up on the shelves for our customers to read and buy. But being super curious about this brand new field, already knowing what it was all about, I spent a lot of my time actually reading all of this content and absorbing it and understanding the hardware and the software and the people and the companies. And the cool thing is that as customers were coming to the store and they would ask the other employees questions about, hey, what does this do? Or how does this work? Or can this board work with this computer? They’d say, I’m not quite sure, but that long haired kid over in the corner, he can probably help you. You had long hair back then? Yeah, I had quite the mess of hair. But the funny thing is my hair doesn’t actually go down. It goes out. And so it’s pretty unruly when it- Yeah, in a fro. Yeah, totally.
Brian Bell (00:02:48):
I’d love to see a picture of Jeff Barrett.
Jeff Barr (00:02:49):
I can dig one up. You can use this as one of your pictures if you’d like.
Brian Bell (00:02:55):
We’ll put it on the thumbnail for the episode.
Jeff Barr (00:02:57):
Yeah. In Amazon terms, we call that diving deep. And so the amazing thing is I’m still doing the same thing today is trying to deeply understand something new and then to explain it to my audience. Yeah. It was super fun to be in the right place at the right time as a teenager. And this little computer store, we had the original, the first personal computer. We had the Altair computer, then the MSI and CompuColor, Southwest Technical Products and the Processor TechSol and all the really originals that are now the retro computers. And it’s actually so fun to see some of my younger colleagues now rediscovering these retro computers and saying, look at this ancient computer I found and realizing, yeah, it’s kind of ancient, but apparently so alive because at That’s something I worked on early in my career. Amazing. And so I did that. Let’s see, moved to Maryland, got married. We had five children, worked in some startups, worked in some bigger companies, and was always doing something pretty hardcore tech and development related. Always loved to be deep in code and building things. Was a VP of engineering for a startup for quite some time, doing cross-platform software. And in 1997, we moved out to Seattle for what we thought would be a kind of a five-year excursion for me to work at Microsoft. Came to Microsoft and I was on the Visual Basic team. Loved the product, had a lot of trouble adapting to being part of a really big company. Decided that wasn’t quite for me. And this was around the year 2000 when web services were starting to become a phrase. And people were talking about SOAP and XSLT and XML and UDDI. and Wizzle, all these really complicated, super early terms for web services. And so for a couple of years, I had a consulting practice that was just all about different customers building and trying to get things to do with web services. The challenge for web services back in 2000, 2001, they were very appealing to a technical audience. And you could show a developer, here’s my server, here’s my client, we’ve got access across port 80 on the internet between the two. Developers are like, wow, that’s amazing. That’s super, super cool. You’d show that to a business person. They’re like, I don’t see why this is anything I should care about at all. And so I was in these technical roles, kind of consulting CTO at a couple different web services, startups around then. And one day I happened to log into my Amazon Associates account. This little box pops up and it says, Amazon now supports XML. I’m like, I know what that’s all about. And this was one of those, I call this like the long chain of improbable events. I see this little box and I immediately click through to see what it’s all about. And I see that Amazon has just launched this new set of APIs to give developers access to the Amazon product catalog. I register for it right away. This is the spring of 2002. So 23 years ago at this point.
Brian Bell (00:13:09):
That’s amazing. What I love about your whole background and story is sort of the... Yeah, it’s serendipitous and there’s lots of unfortunate or not fortunate, but random strings of events, right? And it’s kind of who you are since you were a teenager, right? You found this thing that you were doing working in the computer store, talking to people about the hardware and the software and what you can do with it. And you’re kind of nerding out and you kind of end up and you get a master’s in communication, digital media as well. And so you’re kind of marrying this like communication ability and with this love of technology. And it’s like the perfect evangelist, right? And you kind of found the perfect role for you personally that goes all the way back to the kid with the fro in the Green Lake computer store.
Jeff Barr (00:13:48):
Exactly. It does. And I think what I figured out is when there’s something that I find useful or interesting, I almost just can’t stop talking about it. I just find it like, okay, if I like it, perhaps someone else will too. Never to the point of being rude about it, but I just always happen to name drop something. And if someone says, hey, that’s kind of cool. What’s that all about? Then I’m like, oh, let me tell you a little bit more about this particular topic.
Brian Bell (00:14:09):
And, uh, I love this career advice too, which is the lesson here. If anyone’s young in their twenties and listening is just keep just working on interesting stuff that you find interesting. And like, it’ll be hard to connect the dots as Steve Jobs would say, looking forward. But as you look back at your career, you’ll be like, okay, all the things things I did kind of led to what I’m doing. And I kind of feel the same way about my venture capital career. I didn’t become a VC until I was really like 40, 41. And it’s all that 20 years of just random stuff I did where I was just following my interests, following up. This is more interesting. I’ll go work on that. And this is more interesting. I’ll go work on that. And it’s kind of led me to do what I do. But if it wasn’t for those dozens and dozens of experiences and jobs and careers, I wouldn’t be the VC I am today. It’s kind of like connecting the dots in reverse is much easier.
Jeff Barr (00:14:54):
Totally, totally agree. And sometimes you’ll meet folks and they’ll say how they saw the future and how the world would be 20 years later, and they carefully plotted a path to get there. And I’m sure some people actually do, but for most folks, I think you look back and say, okay, I invented the fact that it was really obvious long, long after it was clear that it was. And so I actually have given a presentation a couple of times called Create Your Own Luck. I’ve given to some high school students and some others where-
Brian Bell (00:15:22):
I think I watched that one. I think I watched that back when I was kind of struggling. How long ago did you first give that?
Jeff Barr (00:15:27):
Probably like close to five years ago, actually.
Brian Bell (00:15:29):
Okay. Yeah, I should definitely look that up. I think I watched it. A lot of good lessons. What are some key lessons and themes that you’d love to share?
Jeff Barr (00:15:36):
the thing I found is that you can sometimes increase the probability of certain kinds of things happening. And like you might want to meet a certain person, because you want to do something for them or with them or ask for something. And so you can arrange things so that the chance of that goes up. But you also have to make sure that at the point when that meeting happens, you recognize that you’re now at this utterly unique point in the history of the universe. And like, this is your only chance. And you have to actually be super bold and make sure you know what your ask actually is. And the illustration I give in my talk is that back when Microsoft was a tiny company, back even before Windows, they had this beta operating system called Xenix that was their Unix derivative. And I was in Maryland at the time, and the company I worked for, we really needed a copy of Xenix in order to put our product on it early. And I flew out to Seattle with no plan, but with the goal of getting access to a copy of Xenix. So I got to Seattle on a Saturday or Sunday and I’m driving past the Microsoft campus and I see Bill Gates getting into his car. And this is 1980 something. So Bill’s not as well known and there’s no security. And I’m like, that’s interesting. I’ll just follow him. And so I followed Bill to McDonald’s in Bellevue. And I’m literally thinking this is my only chance ever to make this request. And I wait in line and step up behind Bill and he sits down. I sit down and introduce myself and say, hey, I’m Jeff. I’m with this company. And we’d really love to get a copy of Xenix 386 for this new product. And he says, you know what? That sounds cool. Tomorrow, call up, I think it was Paul Moritz. I think he said, call Paul Moritz, tell him I sent you. He’ll get you all set up. And sure enough, that worked. And I called the CEO at the time and said, Jay, you’d never guess what I just did. He’s like, Jeff, you’re my hero forever. I cannot believe you actually did that. And so that was just one of those cool things. But you realize there’s only one point ever you get to make that request. And so part of creating your luck is just doing little things to increase your odds to be in that situation, but then recognize you’re in it and then have your ask just ready to go. And then being super bold and just going for it. And never being pushy or rude or anything, but like, you know, just be human to human and just go for it.
Brian Bell (00:17:48):
Yeah, I love that. What a good lesson. So AWS, I think I’ve heard of that. So it started growing pretty fast. How did your role evolve and responsibilities that you picked up that you never expected over time?
Jeff Barr (00:17:58):
Sure. So at the point when Andy Jassy came in and started writing his famous narrative where he talked about how he’s going to build all these different services. And I was in the same hallway as Andy. So there was Andy and then his executive assistant. And then my office was next. And so as Andy’s writing his narratives, we have one on ones and I get to meet with him and he would give me the latest draft.
Brian Bell (00:18:18):
This was like a PR FAQ even back then.
Jeff Barr (00:18:20):
So the narratives are six page docs that usually used to launch a business. And the PRFAQ is often used more for launching an individual service.
Brian Bell (00:18:28):
Gotcha.
Jeff Barr (00:18:29):
So he’s working this narrative and he often would turn to me for some feedback. And as I’m reading this, and at this point, we really don’t have what you could think of as a marketing function inside of our team. And at that point for Amazon, marketing was a little bit of a, you didn’t use that word without a lot of care because the company really had tried traditional marketing and just was not a good way to market amazon.com. So I’m seeing all these great plans for storage and database and compute and networking. And I’m realizing I’m effectively the only person that’s doing what is at that point, developer marketing, going out and speaking to developers. And I can see his headcount allocation and I don’t see that there’s really a developer relations function. So I wrote a doc for him to share what developer relations was all about. And then one aspect of that was I said, I wanna use a blog to launch some of these new services. And it was really challenging to get the point across of what did I mean by a blog for launches? Because at this point, 2003, 2004, blogs were more personal. They didn’t come from companies. They were more about self-expression than product launches. So I didn’t have any examples I could point to and say, I want to do one like this. But luckily, Andy was very, very accommodating and generous and said, okay, well, just go for it. Let’s kind of see how it turns out. And that little first step of a blog I published in November 2004. I ended up doing that for 20 years and writing a million and a half words before at the exact 20 year mark. I said, you know what? 20 years of blogging and 3,300 blog posts for all these launches. That’s like 30 books.
Brian Bell (00:20:03):
That’s like 30 books. How many services do you think you’ve covered in your career at this point? I mean, all of them. I would just get 150 plus. Yeah. That’s wild.
Jeff Barr (00:20:12):
It was super amazing to do that. It was really fun because, again, I got to do what I did back at the computer store. I got to dive deep into a pre-launch service. I got to understand it. I really, really figured out how to set the voice. It was always coming from me directly to developers. It wasn’t coming from the company. It wasn’t coming from the service team. I would always directly use the service, and I would always make sure that the team would give me direct access to the service while it was under development. I’d always say, look, I love you guys all. You’re doing amazing stuff. I don’t believe a word you say until I can actually use these services for myself. And that turned out to be that the fact that I would always ask for access, the fact that I would use the service, I’d give them feedback. I’d ask really hard questions. Turned out to be a really good way to show that you are serious about deeply understanding it, getting all the details exactly right, making sure that what I write and convey to the customer is precisely what they built. And it isn’t the airbrush marketing version. It isn’t sugar-coated. This is exactly what it is. I’m going to tell the customers, here’s all the cool things it does. Show them how to use it. If there’s a couple of things we’re still working on, I’m not going to leave those as mysteries. I’m going to say, okay, you know, we got to this point, there’s a few things we’re still going to work to get built for you for the future as well.
Jeff Barr (00:21:25):
Very, very quickly, the blog took off and it went from being just a fraction of my job to 70 or 80%. And it probably took 10 years or so to do that. But with very understanding management and everybody realizing it’s a different kind of a work product, it didn’t have to go through the usual amount of legal review and PR review that every other piece of public content had to go through at the time. And they basically said, look, here’s the rules. We expect you to follow them. We trust you. Go for it. That trust and being able to never bend the rules, but to figure out where the guardrails are myself and to say, okay, I’m not here to cause trouble. I’m not here to make somebody’s life difficult. I’m here to really accurately represent what our teams have just sweated bullets to. to bring to life for our customers worked incredibly well.
Jeff Barr (00:22:09):
It was really, really, I always insisted on the direct access, on the accuracy, never tried to make the posts overly complicated, always tried to kind of be kind of for a, not a general audience, but a general technical audience, explain a little bit about the particular area, but don’t speak down to people and don’t make it too hard for people to Don’t overload it with jargon. One interesting thing I developed over the years was sometimes you can kind of lapse into enterprise speak without knowing it. And to have like a built-in like enterprise speak detector in my head and just like kick myself and say, nope, nope, nope. That’s not how regular people talk. And just like forget it.
Brian Bell (00:22:45):
Cancel it, delete it.
Jeff Barr (00:22:46):
Yeah, talking to developers. And because sometimes you just like think, okay, now I’m like, we’re big and successful and this is enterprise product. Use enterprise vocabulary. It’s like, but I don’t work that way. I love to talk to developers.
Brian Bell (00:22:58):
Yeah. And that’s probably part of the reason AWS was so successful, right? It was so accessible for developers, so easy to approach. You know, you had all the code samples and it was just easy to get going. And I think a big part of that was probably your evangelism over the last couple of decades.
Jeff Barr (00:23:12):
I hope so. And so over the years, then the blog grew to the point where we had an entire team that was, we were all working together and I was so comfortable a little over a year ago. I said, you know, I’m We’re reaching the 20-year mark. We’ve got a great team, distributed global team. Almost every one of the current bloggers lives outside the US and English is not their first language. So it was a wonderful team to be a part of and to work with. And after wrapping that up in November, 2024, I spent all of 2025 actually traveling the world and speaking to developers. I managed to hit 14 countries this year. So a personal record I don’t really feel like beating next year for sure.
Brian Bell (00:23:49):
Yeah. Yeah. I’ve been on the road like that too. As a young man traveling, backpacking. After about six, eight, nine months, you’re like, I need a home. I just want to be at home for a while.
Jeff Barr (00:24:01):
Exactly.
Brian Bell (00:24:01):
Looking back, is there a favorite service launch or a favorite blog post that you have? Or maybe you could take those one at a time.
Jeff Barr (00:24:07):
I often just took them one at a time, and I tried really hard not to play favorites with the service teams because if you pick one and say, this is my absolute favorite service team, then all the others are like, hey, Jeff, what about us? We thought you liked us as well. It was honestly, every time I would put something out there, I’d say... I need to do at least as well as I did in the past. And one of the most fun things is once you start to get well-known and people recognize you is people will just come up and say, hey, we really enjoy your content. It’s really been helpful to us. It’s helped me to learn. It’s helped me to progress in my career. I always would... kind of capture those moments. And those will be my customers. And at that point, as I’m working on blog posts and plus one, I think, you know, I met these folks. They liked what I did. I can’t let them down. I need to better myself. I need to at least do as well as I did on the last one, but hopefully even a little bit better. So that direct customer contact and feedback was also a great way just to keep me on my toes and make sure I did the best job I could ever possibly do.
Brian Bell (00:25:08):
So you mentioned all the way till 2024. What happened in 2024? You’re in a new role now?
Jeff Barr (00:25:13):
Yeah, so I still have the role, but now I’m focusing on just traveling and speaking to developers. So in 2024, one thing I realized was that with all of these AI-powered coding assistants coming out, there was a lot of lack of understanding of where they came from and a lack of understanding of what they did. And what it meant for developers. So early in the year, I put together a presentation that I’ve been updating ever since trying to say, here’s the basically like, as developers, we are tool builders, starting from the earliest editors and assemblers and compilers and debuggers and source code control and IDEs, and then say that the AI coding assistants, they’re just really a next logical step along the way. And that I’ve gotten great feedback from developers. That’s a bit reassuring. But I’ve also been focusing on making sure that developers understanding that there’s new skills they need to learn.
Jeff Barr (00:26:07):
And one of the main messages is that you need to be as good at reading code as writing code. And I think our schools have done a great job of teaching developers how to write code, but often from a blank screen or a blank page. There hasn’t been a curriculum as far as I know that says, here’s a massive pile of somebody else’s code, figure out what’s inside there and make sure that you know how it works and what it does and review it for security issues and for correctness.
Brian Bell (00:26:33):
We actually have a startup that does that. I’ll send you their name after. Maybe take a look at them. But basically, they’ll crawl the whole enterprise code base and help. Basically, that’s the use case is helping new developers on the code base get up to speed, right? What are all the conventions here? What’s all the logic here? And how do we program these functions and so forth?
Jeff Barr (00:26:52):
Exactly. So my thinking was that when developers are using the coding assistants, the coding assistant is going to write the code. The developer needs to look at that code and say, does this code actually reflect what I asked and what I want to do? Because once that developer actually checks that code in, it’s the developer’s name on the commit. It’s not the coding assistant’s name. So the developer is ultimately still responsible. So I say reading code is now an essential skill. And I also have said that being a really good, effectively human to human communicator, is the other skill that developers need to do. So as developers, we’re often taught to write code. And we learn the language, we learn the syntax, the data types, the functions, the operators, the libraries. And we’re like, okay, great, we know it. We can put it to use in a bunch of different ways. But when we’re talking to our coding assistants, we’re not writing code, we’re writing hopefully accurate, descriptive, literate, human readable text. And not every developer is as good at that as they could be.
Brian Bell (00:27:48):
Yeah, I mean, you’ve lived through cloud now, the scale up of mobile, SaaS 2.0. How does this AI wave feel to you compared to past paradigm shifts?
Jeff Barr (00:27:58):
Well, the interesting thing about a lot of these paradigm shifts is at first, it feels like there’s just a lot of hype and everybody gets super excited. And then the conservative party says, no, this is never going to work and it’s never going to do the right things. But then you start to see people getting real value from it. And you start to see the, not just getting real value, but people really coming out with entirely new ways to put it to use that are surprising. And what I find fascinating is there’s some really cool GitHub repos where they’ve extracted the system prompts from a number of different AI coding tools and other AI tools. And to read these system prompts, you realize that there’s a whole different skill involved in writing system prompts than writing, I don’t know, a C compiler or an operating system or something. So I’m not saying that traditional developers can’t write prompts and get used to them, but there’s a different set of skills that are, you’re generally a little bit more verbose or you’re writing in terms of should and can and must and must not versus the kind of the traditional logic.
Jeff Barr (00:29:00):
And one of the things that is actually really hard to get used to with the new tools is developers, we strive for like accuracy, like every piece of punctuation counts and the spacing counts and like make it pretty and not a single typo. Which is, it’s still good to express yourself accurately. But when you’re kind of using these AI tools in a quick and dirty way, you almost have to train yourself. I make a little typo. The AI is still going to figure this out because many other people have made that same typo anyway. It’s not worth the time sometimes to backspace and fix it. You might as well just keep on going and express yourself quickly and a little bit messily versus being precise. And that’s a hard habit to unlearn.
Brian Bell (00:29:39):
Yeah. You know, I think, you know, I use AI in everything I do at Team Ignite. And, you know, what I’ve learned to do is just basically press, you know, the voice to text button and just, you know, diarrhea stream of thoughts into the LLM, right? You know, everything, just a stream of consciousness. And then the LLM will kind of figure it out based on that stream of consciousness. You know, just try to get all my ideas out. And you can talk, you know, you know, three to four X faster. then you can type, right? And so you can just get all your thoughts out and just let it just get it all into the context.
Jeff Barr (00:30:10):
Exactly right. And so my programming in college, so I started out in a community college and I was probably like the last generation to learn on mainframes with punch cards. And so with punch cards, that accuracy is paramount because you’re actually like making holes in a physical card. But the turnaround time between writing the code and getting the either successful compile and run or the errors back might be four, eight or 12 hours. And so desk checking, as we used to call it, and making sure everything was perfect before you submitted your job would save you. You’re either going to spend a few minutes checking beforehand, or you might lose an entire day’s worth of progress. And so that mindset of perfection and checking ahead of time, which still seemed like great skills to have, voluntarily setting those aside and saying, I know I could be, but I don’t need to be. It’s a really interesting kind of...
Jeff Barr (00:31:00):
It’s the same thing with our computer keyboards versus our phone keyboards. The phones are usually really good at auto-correcting. And you talk to them, you put in approximately what you want, and it corrects nicely to the point where you, what you did need. But our computers don’t always do the same thing. So sometimes the simple device is better than the complex one.
Brian Bell (00:31:20):
You guys just had reInvent. Usually it’s the week after Thanksgiving. Unfortunately, yes, it’s always the week after Thanksgiving. I remember I launched SageMaker Marketplace back in 2018. And I basically worked all through the Thanksgiving week trying to get that thing out. How would you, you know, let’s start from a, you know, like a bird’s eye view or a top level view. What are your key takeaways? And then we can kind of go into each of the, all the things you guys announced.
Jeff Barr (00:31:43):
Yeah. ReInvent gets better every year, it seems. And one of the things that really struck me this year is, yes, we did a bunch of amazing launches and we pushed forward in a whole bunch of different frontiers. There was a lot of just community and a lot of connection happening. And it wasn’t just me. We actually got some great feedback from a lot of industry analysts that said something that’s clearly setting us apart is the size and the strength of our communities.
Jeff Barr (00:32:13):
A lot of what I’ve been traveling to this past year has been to our communities in South America, Central America and Asia and Europe and going to visit all these communities and seeing how strong, how passionate they are. And these communities are not built by AWS. They’re not sponsored or supported by us. These are basically just users that get together and say, we so much see the value of tech and cloud AWS for ourselves and our community and our future. We’re going to form a user group and we’re going to get together. We’re going to meet. We’re going to learn. Communities all over the world create these things called community days that are in many cases in the multi-thousands of attendees. And so to get to meet many of the AWS community members at reInvent, it’s always, some of these are like meeting for the first time, but a lot of them I’ve met in Chile or Argentina or Peru or Bosnia or Switzerland or all the cool places I’ve gotten the chance to visit this past year, but to see that strength of the community, and we do have this community builder recognition program to help these community builders learn from each other. That community strength was really, really apparent this year. The work that the community is doing to create and push themselves forward is super, super impressive.
Jeff Barr (00:33:21):
That was great to hear just the breadth of announcements, everything from the ground on up, from the servers to the serverless, to the storage, to the AI. I really strongly think we had something for just about everybody. And if we didn’t have a chance to announce it, it was almost not because we didn’t have anything. It was because time and space are finite and we can’t do everything all at once.
Brian Bell (00:33:44):
Yeah, I remember Jesse’s slides on, you know, like basically the services are going up into the right. How many services are in AWS now? Like there’s got to be over a thousand, right?
Jeff Barr (00:33:52):
No, we’re in the high tier, I believe. And we’re definitely aware that there is a challenge of when do you build new services versus when do you add new effectively things to existing services? Trying to balance that out and keeping all the naming consistent and clear is always something that we’re struggling for.
Brian Bell (00:34:10):
Back to the network thing, the grassroots community-based aspects of AWS, that definitely resonates with me because, you know, both of us have worked at Amazon and Microsoft and, you know, I went to Microsoft after AWS and that was kind of the culture shock for me when I went. You know, because AWS always felt like kind of flatter and more nimbler and more grassroots, you know, more developer focused. And then Microsoft always felt like a little bit more corporate and top down. And, you know, these are generalities. I mean, there’s definitely pockets of both of both companies, but I don’t know what your perspective is.
Jeff Barr (00:34:43):
That large degree, although my Microsoft experience is now 25 years old, so I can’t speak to how it is now. But what we do at Amazon and really making sure the community understands that they know that we know that they’re doing something important. And so we’ve got this community builder recognition. I visit the community days all over the world and I speak there. And also what we often do whenever I’m traveling to a new location, I will message the local AWS office and say, you know, I’d love to meet some customers and some community members while I’m there. And so quite often we’ll do something like a round table event and they’ll invite the leaders of the different local communities. We’ll get together. And the stories that they tell you of themselves and the members of the growth and the change in places with Central America, South America, these emerging economies. And they’ll tell you these amazing stories where they’ll say, okay, we had our regular life, our regular job. And we said, we see this brighter future for ourselves in tech and very specifically in AWS.
Jeff Barr (00:35:46):
And they They worked so hard to upskill and reskill themselves and to qualify for new jobs. And it’s not like those of us here in Seattle where we’re kind of in a very privileged position. Okay, we get a new job and we get our 10 or 20% boost. Like, oh, wow, that’s awesome. The folks in these countries, like, we got a 10x difference in our pre-tech life to our current life. And it changes everything. And it literally brings you to tears when you hear, okay, we were... At this income level, we applied ourselves, we worked so hard. We got our first AWS certification. We got a job as a solutions architect. We are now, and the lives of themselves, their children, their parents, everything has changed for the better. And that is such an unappreciated value of the community here because they’re not doing it by themselves. They’re attending the events. They’re learning at the events. They’re networking. They are forming... what you kind of call like a mutual support group within these.
Jeff Barr (00:36:42):
And they say, okay, let’s all get together and we’re gonna spend a day every week studying for our next exam. And we’re gonna all, we’re gonna take it. We’re going to figure out how we’re gonna partition up all the study materials. We’re gonna review together. We’re gonna quiz ourselves. And they’re doing it under not the best of situations. They’re doing it while they’re raising their family, while they’re working their existing job. They might be, maybe English isn’t their first language. So they’re also working to translate not just the content, but the names of the services. And the names of the services get sometimes translated, sometimes not. That adds sometimes confusion and messiness. So you walk away from these meetings with just nothing but respect and awe for everyone. Not for how hard they worked and the fact that they actually like, they saw something brighter for themselves and they went and did it. And it just, it’s always just like, it just brings you to tears of like, wow, I’m so happy for you that you made this amazing change and that it’s going to be a, and like we know generation after generation, right? You lift up one generation, the successive ones are also going to be lifted up. And to know that the little bit I did to help that was valued and recognized is always great to hear as well.
Brian Bell (00:37:48):
Yeah, it’s sort of the rising tide of technology. Starting with the internet, accelerated by mobile and cloud, and now even further accelerated by AI.
Brian Bell (00:37:52):
You can be some kid in Africa with a $100 Chromebook and start studying how the cloud works. Get AWS certified or whatever certification and get online and start 10x-ing your income or 50x-ing your income. from some country most of us have never heard of in Africa. And that’s just, I love that story because it’s really the story of kind of America, the American dream in a way. We’re almost like spreading the American dream around the world.
Jeff Barr (00:38:27):
I sure hope so. So I was in Bangalore earlier this year and being interviewed by some journalists.
Brian Bell (00:38:32):
Amazing tech community there.
Jeff Barr (00:38:33):
Amazing. It totally was. It was great to spend time with them. And I was talking to some journalists and kind of on the spur of the moment, we came up with this phrase that the uni unicorn, that the unicorn with one developer at the heart. And with the way AI dev tools are going and with all these powerful cloud services, it sure seems like the opportunity is going to be there for one developer to build a unicorn at some point. It’s almost all the trends are pointing toward that.
Brian Bell (00:38:58):
Sam Altman and his famous, like the text messages got leaked. They famously have a bet over the over under, I think is 2030 for the solo entrepreneur unicorn.
Brian Bell (00:39:08):
Yeah, exactly. Which means that person’s probably maybe already working on the problem now solo and working on the solution now and starting to charge revenue for it now because it usually takes... you know, five to seven years to become a unicorn. So that’s the, and I see it as an early stage investor. I see, you know, two founders get to a million of revenue before they hire anybody.
Jeff Barr (00:39:26):
I think the tools and the opportunity there, and we need someone like as usual, see a cool, a cool unmet need and just sit down and build it. And the time is better than ever to do that.
Brian Bell (00:39:35):
Yeah, that’s so exciting. You get to watch that whole shift. Back to reInvent, what were some of the AI, agentic, Bedrock kind of things that you guys announced?
Jeff Barr (00:39:44):
Well, let’s see. So I know that we announced a bunch of new models for Bedrock. I don’t even know the number. It was like 20 or 25 new models. A lot of add-ons to make it even easier for developers to build agentic apps. One of the interesting things is developers love to come up to me and show off what they’re doing, but usually they just talk about it. But for some interesting reason, within the space of about four hours one day, I had three separate developers come up to me with their laptops and show me their agentic apps they were building. And it was like the most interesting coincidence that I’ve ever seen.
Brian Bell (00:40:13):
Yeah, that is such a fascinating thing. And I think I’m definitely seeing that as an early stage investor as well.
Jeff Barr (00:40:19):
Yeah, a lot of them make use of our strands SDK. It makes it really easy to build agentic apps.
Brian Bell (00:40:24):
That’s awesome.
Jeff Barr (00:40:24):
Good to just see the diversity of the models that we’re doing. It’s fascinating how quickly these models emerge. And you kind of think like each of these models are very complex to architect and to build and to train and to test. You kind of like, I’m always a person who whenever I see something, I’m always thinking, how did it get here? And what was behind the scenes? And so you’re kind of thinking like the entire world is just like, there’s all the smartest people are working on building these models now. And yeah, pretty impressive.
Brian Bell (00:40:52):
Well, and you think about agentic, frameworks and AI, it’s enabled by all the stuff that AWS and others have built over the last 15, 20 or decades, really. You can connect it all the way back to the earliest services at AWS. And then you just kind of keep building abstractions.
Jeff Barr (00:41:09):
That is the thing that is, that sense of history, I always like to bring to what I’m doing. And the fact that I’ve, one of the fun things about a career in tech is that within decades, you get to see so much evolution and so much change and so much growth. And you get to see all the layers of technology actually each emerge and each one emerges. And there’s the early adopters and there’s the skeptics and the reality is always somewhere right there in the middle. And then it goes from like, everybody’s skeptical to like, it’s now this just accepted part.
Jeff Barr (00:41:38):
of how things work. And so layer by layer, you see all these neat pieces of tech accumulate. And there is this kind of painful difference to me with the models is that every other layer of tech, it was pretty easy to take it apart and understand what was inside. Because as developers, we like to build abstractions. We write to those abstractions. But every so often, we want to kind of lift the covers and look inside the abstractions for performance or debugging or just sheer curiosity. And we could do that with our operating systems, our editors, our compilers, all those tools. Maybe I’m just so disconnected from the way the models work that there is a way to do it, but it is fundamentally different is my sense. There is no easy like view source where you can look inside and say, okay, this is what’s truly going on inside one of these models.
Brian Bell (00:42:25):
Right, it’s all statistical now rather than deterministic, right? So you kind of have to, it’s almost like you would manage humans, right? And you get variable outcomes, right?
Jeff Barr (00:42:35):
We’re all statistical at some point, right?
Brian Bell (00:42:37):
Yeah, fundamentally, we’re massively parallel, right? What an interesting transition. What about the infrastructure stuff? You guys announced new chips and networking and server architectures. I’m sure you got some new databases.
Jeff Barr (00:42:50):
Yeah, from the Tranium 3. And the Tranium 3, and I actually was just watching one of the keynotes earlier today. The Tranium 3, like each Tranium 3 is incredibly powerful. powerful, but you put like, I think you put four of those on a board with a Graviton and with the security chips, the Nitro chips, you put two full racks of those and you get the Tranium Ultra server. And then you make a whole data center for those, you call it the Ultra cluster. And you start multiplying all the numbers of how many and how big and how fast. You get to these numbers, you can’t match to any kind of reality at all. We can understand hundreds and millions and billions. When we get to trillions, it’s hard to match that to any kind of a reality, right?
Jeff Barr (00:43:34):
There’s not that many things we can look at and say, okay, I can mentally grasp a trillion of something. But trillions are kind of like,
Brian Bell (00:43:41):
Yeah, you start getting into the end notation, like 10 to the 14th, and now it’s 10 to the 18th. Okay, and now it’s 10 to the 22nd. Okay, like it’s hard to like, we’re not good at like fathoming orders of magnitude like that.
Jeff Barr (00:43:54):
Exactly right. Yeah. And so just unimaginable amounts of compute power that are there just waiting for clever developers to figure out cool ways to take advantage of it.
Brian Bell (00:44:03):
What are you excited about over the next five years? We’re sitting here right at the end of 2025. What are you excited about coming up?
Jeff Barr (00:44:09):
So I suspect that we’re going to see a new kind of developer emerge. And some of these are going to be existing developers evolving themselves into this new form. But I think the new kind of developer is going to have a massive amount of understanding and empathy for their users and their customers, and then just an unparalleled ability to translate that into requirements and specs and say that in a very descriptive, lucid, accurate way, say that this is the problem that I want to solve for my customers. And so I think we’re going to start to see developers that can do that successfully be outrageously successful and outrageously productive.
Brian Bell (00:44:46):
Yeah, I heard from, I think it was one of my founders. He said that the role of the product manager and the developer is kind of merging in a way, right? You don’t, yeah, you used to have to like, you know, here’s the specs, go build this. And okay, here’s the thing, go test it with customers. But now it’s kind of like merging into one person.
Jeff Barr (00:45:05):
I think the model of the app itself is going to change too, where we’re going to have a set of apps that we’re going to see a problem, write our spec, build the app, use the app, and we’ll probably keep the spec around for the next time that we need the app. But we might end up with this kind of part of the pie chart is effectively just disposable, discardable apps where it’s just not worth the trouble to keep them around for the long term because the tools are going to keep getting better.
Jeff Barr (00:45:31):
Our work to understand what the user’s needs are and to put that into a requirement stock and a spec, that’ll be our long-term value. But the artifact of this finished app, maybe it’s not worth keeping around and maintaining for the long term. It used to be we’d invest so much upfront to get this app, figure out the problem space, spec out the app, build and test and deliver the app. The app was then really precious and curated. We need to keep it around forever. Maybe that’s not true anymore. We just build them, use them and toss them.
Brian Bell (00:45:57):
Yeah, kind of like the, I think Sam Altman calls this the fast fashion era of SaaS.
Jeff Barr (00:46:03):
Yeah.
Brian Bell (00:46:04):
And I heard that on my last episode.
Jeff Barr (00:46:05):
Hopefully there’s less natural resources wasted, but that’s a different challenge.
Brian Bell (00:46:09):
Yeah, yeah, yeah. Yeah, we’re not filling up landfills with clothing.
Jeff Barr (00:46:12):
Exactly. Yeah, but this idea of this, like there’ll be the durable apps that are effectively like the EC2s and S3s and DynamoDBs, those are going to be built and they’re going to last for decades and centuries.
Brian Bell (00:46:24):
Yeah, those are still the kind of discrete units of compute and infrastructure that, you know, but then there’s just this abstraction layer that’s whipped up the stack where you can, yeah, just talk to the AI.
Brian Bell (00:46:28):
Hey, I need this thing. And it kind of classifies in this way. And then it updates and I need to be notified in that way. And, you know, and so now you can just talk to, you know, effectively the infrastructure, the database, rather than having to interact with like a SAS layer.
Jeff Barr (00:46:49):
Exactly. Build it and use it. And at some point, it’s just too much trouble to actually give it a name and keep it for later. Because the next time you need it, you’ll say, okay, I need an app to do the following. And you can either search for it or your search is like half of a spec anyways. You find the spec and go for it. So I think we’re going to get to that point for certain kinds of apps.
Jeff Barr (00:47:09):
On the flip side, I do think that... It’s going to be so easy to build apps that the long-term preservation will be the data. Having awesome data warehouses that are well-organized and well-understood and well-labeled and schematized and super easy to use, I think all these cool apps we can build so quickly are going to say that having great data is going to be an even more powerful advantage.
Brian Bell (00:47:32):
Yeah, love that. Let’s wrap up with some rapid-fire questions.
Brian Bell (00:47:35):
Sure.
Brian Bell (00:47:36):
What’s a recent AWS announcement product or service that most surprised you?
Jeff Barr (00:47:40):
Well, let’s see. I think the durable Lambda functions are super cool. This idea that the functions can stick around for up to a year and they can run for a while and they say, hey, let’s hang on for some external input. And the function just pauses in a durable way and infrastructure might change. And then the thing happens and it keeps on going. I really didn’t know that was a thing we were working on. So that was pretty cool to see that pop out.
Brian Bell (00:48:05):
Is there a misconception about AWS or cloud broadly that you’d like to see disappear in the next year?
Jeff Barr (00:48:11):
The old misconception was security. And I think we’ve addressed that with really good features and user education. What’s a misconception? I I can’t think of one that I hear again and again where I’m like, oh my gosh, I thought we addressed that. There was a time early in AWS where security would come up every time. And there’d always be like the bearded old guy at the back of the room would say, this sounds amazing, but have you thought about security? It’s like, yes, we have thought about this really deeply. Let me tell you all the ways that we actually have. And that hasn’t been the case for a decade or so.
Jeff Barr (00:48:43):
So there was that, and the original misconception, which we should have stumped out many years ago, was that the original AWS was simply about extra capacity in the data center. That was one of those things that the first time we saw it, we should have said, this is not true, but we just didn’t actually address that. We’re going back 15 plus years at this point, but we successfully addressed that many times, but that was fun. That was one of those things that over and over again, you have to keep correcting people’s misunderstanding on.
Brian Bell (00:49:08):
What are you building or exploring personally right now? Side projects, experiments that are exciting?
Jeff Barr (00:49:14):
Let’s see. So I’ve got way too many side projects. I’ve been doing a lot of fun 3D printing of 3D printing fabric and geometric objects and designs. Having lots of fun with that. I’ve got five different Raspberry Pis running in my home office here. I have a magic mirror screen. I’ve got Raspberry Pis running my 3D printers. I’ve got one running that fake PDP-11 behind me. I’ve got another one over there that’s listening for flight data and sending that up to FlightAware. So I’ve always got a lot of cool stuff going on. And I always like to be directly connected to things. I’ve successfully resisted the urge to kind of abstract myself and only talk in enterprise generalities and make sure I stay firmly connected to like, feet on the ground, can still take a Raspberry Pi and get it operational and still dive deep into the code and figure out why something’s working or not working. And I just never want to lose that ability to go hands-on and make stuff work and fix it if it doesn’t.
Brian Bell (00:50:10):
Creates authenticity with the developer community as well.
Jeff Barr (00:50:13):
It totally does. And there’s a certain sense of like, I know how things work. And there’s nothing that’s just this vagueness beneath me that I can’t, with enough energy, figure it out.
Brian Bell (00:50:23):
If you’re starting over today, fresh grad, first job, what stack or domain would you dive into?
Jeff Barr (00:50:28):
Man, stack or domain. I’m still really excited by everything serverless. I still think that serverless in general is a super, super cool concept. And I think it’s still getting started. The power to just spin up this, hand your code over to the computer and it does all the rest for you. And that’s almost taken for granted, but not quite yet.
Brian Bell (00:50:50):
I remember when it came out, it was like, wow, this is amazing. And now it’s just kind of taken for granted. And I think the agentic world is going to definitely accelerate that even more, right? Kind of really rely on that.
Jeff Barr (00:50:59):
Developers are going to build these agentic apps that have a lot of different moving parts. We’re going to see a generation of developers that maybe don’t know what all those different parts need to be. And they’re just going to say, I built this thing with a bunch of natural language prompts. Why do I need to go into all this deep tech in order to just simply deploy it and get it out there at world scale?
Brian Bell (00:51:19):
Yeah, I just want it to work.
Jeff Barr (00:51:21):
Yeah. Now, there’s going to be this fun conflict between the deep technical folks that those of us who’ve grown up from, okay, we understood semiconductors and Boolean logic all the way up to computers. And we’re like, okay, we’re the authentic old school geeks. And there’s going to be these new folks. They’re like, that’s amazing. And I could care less about that.
Brian Bell (00:51:42):
Yeah, I mean, I think there’s a time and place for the hardware up, you know, like when you really need to build a scalable, repeatable system that services hundreds of millions of people, like maybe then, okay, great. Then you’re going to go through the stack and completely optimize the delivery of those bits and bytes. But, you know, for everyone else, you know, serverless is fine.
Jeff Barr (00:52:01):
It’s like there was kind of this old school where users were held in great disdain, right? There was this kind of like computer priesthood. It was like, oh, those users, they wreck everything. And we’re going to see the same kind of thing where it’s like, oh, those AI people, they just don’t even know what they’re doing. They’re just like solving customer problems all day and night, but they actually don’t know what they’re doing.
Brian Bell (00:52:19):
Which is like what all this stuff exists to do anyway.
Jeff Barr (00:52:21):
Exactly right. To solve people’s problems. There’s always this fun tension between the two.
Brian Bell (00:52:25):
Yeah, that’s... Last question. What do you want your legacy to be?
Jeff Barr (00:52:29):
My legacy? I want people to remember that I gave them some good information and facts that helped them to improve themselves and their career.
Brian Bell (00:52:36):
I love that. Well, I learned so much, Jeff. Thanks for coming on. I want you to come on for a couple of years and I’m glad we finally sat down.
Jeff Barr (00:52:42):
This has been super fun. It’s gotten dark in here as the sun’s been going down in Seattle so early.
Jeff Barr (00:52:47):
Yeah, it reminds me.
Brian Bell (00:52:49):
It gets darker up there earlier than down here in California.
Jeff Barr (00:52:52):
Yeah, what is it? It’s only 4.05. We’ve got like a few more minutes of daylight. And my window to my right, it faces south. And so at the very, very peak of the winter solstice is when I get the sun actually sneaking in the window here. But I had to put the shade down. It’s getting dark really fast.
Brian Bell (00:53:07):
Well, thanks so much, Jeff. I appreciate you coming on.
Jeff Barr (00:53:08):
Been a pleasure.
