What if your finance team was an intelligent swarm of agents—working 24/7, reconciling your books, answering your questions, and evolving every month without adding headcount?
That’s the future Erin Kim is building with Mesh, a startup creating fully agentic, AI-powered finance teams—starting with one of the most painful, repetitive tasks in accounting: bookkeeping reconciliation.
In this episode of Ignite Startups, Erin joins host Brian Bell to share her founder journey from investment banking to Carta to YC-backed founder—and why now is the moment to reimagine the structure of modern finance operations.
🚀 From Investment Banking to AI Automation
Erin’s career began in investment banking at Barclays, covering industrials. But even in the most traditional of financial environments, she gravitated toward automation. “I kept trying to automate myself out of a job,” she jokes. That mindset led her to Carta, where she joined as one of the earliest finance hires—and ultimately transitioned to product, building internal fund accounting tools.
At Carta, she met her future Mesh co-founder, Nandini. Together, they spent years turning recurring fund workflows—capital calls, distributions, reporting—into scalable, automated infrastructure. That experience formed the foundation for Mesh.
🤖 What Is an Agentic Finance Team?
Mesh isn’t just automation—it’s an agentic system. Erin explains that the big shift unlocked by large language models (LLMs) isn’t just faster workflows, but the ability to have back-and-forth conversations with your software. Unlike rigid SaaS workflows, agents can reason, learn from corrections, and operate across systems.
Mesh’s first product? A cash reconciliation agent that pulls transaction data from banks, maps it to your general ledger, and flags discrepancies. But that’s just the beginning. The roadmap includes agents for payroll, invoicing, billing, and more—creating a full-stack finance team powered by AI.
🧠 Trust, Accuracy, and the Human in the Loop
One of the biggest challenges in AI-driven finance is trust. Accounting isn’t a domain where “close enough” is good enough.
Mesh tackles this with a hybrid approach: the AI takes a first pass, classifies transactions, and then pings the user for final review. Each correction is stored in memory, improving accuracy over time. Think of it like training a junior analyst who learns quickly—and never forgets.
Their KPI? Precision. Erin aims for 99%+ reconciliation accuracy—matching or exceeding human performance.
🔄 Replacing Software… and the People Behind It?
The rise of Mesh signals a deeper shift: from Software-as-a-Service to Work-as-a-Service. Instead of giving customers tools to do the job, companies like Mesh do the job for them, with minimal oversight.
This unlocks an even bigger question: Will startups of the future skip hiring finance teams altogether? Erin thinks yes. In her vision, founders might scale all the way to Series B or C before hiring a single in-house finance person—and that person might act more like a system controller managing AI agents.
🌐 Go-to-Market and Early Traction
Mesh is seeing strong inbound demand, particularly from Series A+ companies looking to scale operations without bloating headcount. SMBs and startups with growing transaction volumes are ideal customers—especially those not yet ready to hire a full finance team.
While Erin sees potential partnerships with accounting firms in the future, the current focus is direct adoption by companies that feel the pain acutely.
💡 Surprising Insights and Advice
Erin shares how the dual nature of selling AI—excitement meets skepticism—has shaped their UI and onboarding approach. Building trust is key, and that means showing your work, involving the human, and reducing friction over time.
Her advice to early-stage founders? Start building and sharing early. Waiting for a perfect product is a trap—momentum comes from getting in the game.
🧠 Favorite Tools, Trends, and Philosophies
Outdated finance tool? Blackline—"It’s more workflow than real reconciliation."
Underrated AI use case? Secure agent identity and credential management—think 1Password for AI.
Favorite startup in her YC batch? A1Base, creating digital identities for AI agents.
Most underrated challenge in AI? Memory and recall, especially across systems.
⚡️ Why This Matters
As Erin puts it, we’re entering a world where interfaces are built on the fly, agents do the work, and founders can scale faster with fewer people. Mesh isn’t just making finance easier—it’s redefining how finance works.
For VCs, operators, and startup founders alike, this conversation is a window into how foundational AI agents will reshape core business functions. Not just with flash—but with real-world, verifiable accuracy.
👂🎧 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:
Welcome & Guest Introduction (00:01 – 00:34)
From Investment Banking to Automation (00:35 – 01:22)
Joining Carta and Scaling Strategic Finance (01:23 – 01:53)
Shifting into Product and Fund Accounting (01:54 – 02:48)
Founding Mesh and the Origin Story (02:49 – 03:39)
Building Automated Finance Workflows (03:40 – 04:24)
The Agentic Future of Finance (04:25 – 06:20)
Balancing AI Accuracy and Trust (06:21 – 07:56)
Human-in-the-Loop Workflow at Mesh (07:57 – 09:12)
What Can and Can’t Be Automated (09:13 – 10:10)
The AI Unlock: From Workflows to Agents (10:11 – 11:27)
The UI Shift in the Agentic Era (11:28 – 13:11)
Work-as-a-Service and the Mesh Interface (13:12 – 16:16)
Who Mesh Serves Best (16:17 – 18:05)
Selling AI Tools: Excitement vs Skepticism (18:06 – 19:14)
Long-Term Vision for Agentic Finance Teams (19:15 – 21:00)
Accounting Talent Gaps and Industry Trends (21:00 – 22:01)
Advice to Early-Stage Founders (22:02 – 23:18)
Key Metrics and Accuracy Goals (23:19 – 25:23)
Context Switching and Founder Focus (25:24 – 27:29)
AI’s Untapped Use Case: Secure Credentials (27:30 – 28:12)
Balancing Advice and Intuition as a Founder (28:13 – 28:38)
Share this post