A Product Market Fit Show | Startup Podcast for Founders
He raised $75M—but kept his team to 21 people & pays everyone the same salary. | Cos Nicolaescu, Co-Founder of Accrual
6 Jul · 58 min
Key points
- Accrual co-founder Cos Nicolaescu says the company's first real product-market fit signal came about a year in, when complex-document returns from firms like Armanino and Creative Planning showed staggering savings.
- Nicolaescu says customers asking for adjacent tax returns and audit-side use is the real validation, and Accrual's pilot-to-production conversion rate is 100%.
- Nicolaescu argues that with AI, all-in-one platforms are higher risk and higher reward, and that Accrual keeps 20 people at the same salary with no management layer because coordination costs rise with individual productivity.
Accrual raised $75m and still runs with 21 people
Accrual is building AI for tax and accounting firms, and its early traction comes from customers who pay for hard work and then ask for more. Nicolaescu argues that as AI multiplies individual output, small teams with no management layer and heavy equity compensation outperform large ones. Making AI reliable in a regulated industry takes far more than a demo.
Takeaways
- Paying customers asking for more signals product-market fit. Accrual felt fit about a year in, after its first returns for complex documents at firms like Armanino and Creative Planning showed huge savings. Customers now pay for complex work, request adjacent return types, engagement letters, and audit use.
- All-in-one AI platforms are high risk, high reward. Before Accrual, top firms used three to eight tools, and clients needed at least two portals for tax work. Nicolaescu argues AI needs full workflow context, and says the test is how many tools a firm can deploy live, not pilot.
- Pilots compare output against finalized returns. Accrual reached 100% pilot-to-production conversion by being truthful, iterating daily, and comparing its output side by side with each customer's finalized returns. Nicolaescu criticizes vendors whose pilots use humans to fix AI output.
- Anchor customers and tax partners drove go-to-market. H&R Block, which handles about 20% of American filers, and top 20 firm Armanino were first customers, since scale and regulation force rigor. Accrual shadowed accountants first and targets tax partners, because CEOs sit too far from details.
- Tiny teams beat coordination costs. Accrual has 21 people after raising $75m. Nicolaescu argues coordination costs rise as individual productivity rises. It has no managers, one-on-ones or scheduled meetings, hires experienced people, runs work trials, and rotates people between projects every few months.
- Equal pay and heavy equity attract believers. Everyone earns the same salary, founders earn less, and most hires took small pay cuts. Accrual models equity value over four to five years. Srugo notes a small headcount allows bigger grants than rivals with 50 or 100 staff.
- Industry-wide impact guided the choice of accounting. After months of ideas, the founders listed what they would avoid, including crypto, foundation models and consumer products. They chose accounting for broad impact and because large firms kept asking for help, drawing on Nicolaescu's Stripe and Brex finance experience.