Building the Intelligence Engine for Finance and National Security with Shaun Modi of Capitol AI

Highlights

• Founded Capitol AI in 2021, formalized the business in 2022, and accelerated through Y Combinator’s Summer 2024 cohort — after Airbnb co-founder Brian Chesky personally encouraged him to apply
• Built an early go-to-market around health insurers, then re-architected the entire tech stack around multiple large language models once the LLM wave made the original approach obsolete
• Career path spans RISD industrial design, NASA lunar vehicle design, Motorola, Google (working directly with co-founder Sergey Brin), and Airbnb, where he served as founding product designer
• Advised the White House and Department of Defense on a COVID-era supply chain risk program for NORTHCOM/NORAD — the direct origin point for Capitol AI
• Counts EY as a flagship partner, powering commercial due diligence and divestiture workflows that compress weeks of manual work into half a day
• Took runner-up honors at the American Banker Conference
• Leads a team of 26, with roughly half of Capitol AI’s revenue currently generated in the UK ahead of a planned EU expansion out of London

Summary

Shaun Modi’s path to founding Capitol AI reads like a tour of some of the most consequential product organizations of the last two decades. Trained as an industrial designer at RISD, he designed lunar vehicles for NASA, rugged devices for Motorola, and search and personalization experiences at Google before an acquihire brought him to Airbnb as a founding product designer. That design pedigree, paired with a stint advising the White House and Department of Defense on a COVID-era supply chain risk program, gave him a front-row seat to a specific and painful problem: government analysts producing reports across sixteen disconnected dashboards from sixteen different vendors, with no single architecture tying together AI models, data, and human oversight. That gap became the founding thesis for Capitol AI.

The company didn’t arrive at its current form in a straight line. Capitol AI’s first go-to-market targeted health insurers, and while the team generated real revenue, two forces converged to force a change: the arrival of large language models via API, and the notoriously slow, difficult sales cycle inherent to health insurance. Rather than treat the shift as a failure, Modi frames it as a natural consequence of operating with limited information — a premise every founder starts with, tested and revised once real customers start talking back. The company re-architected around a flexible, model-agnostic stack (now spanning more than 500 models) and used that flexibility to find where revenue and value actually lived: financial services, professional services, banking compliance, private equity, and government national security.

That strategy has paid off. EY is now a flagship customer, using Capitol AI to power commercial due diligence and divestiture workflows that once consumed weeks and now take half a day — always with a human reviewing the output. From there, Capitol AI found adjacent markets in regulatory compliance for banks, portfolio reporting for private equity funds, and procurement acceleration for national security agencies evaluating vendors and technology performance. The team recently took runner-up honors at the American Banker Conference, evidence that the pivot from healthcare paid dividends.

Much of the conversation also explored what AI does and doesn’t replace — a natural extension of Modi’s design background. His view is unambiguous: AI is a tool, not a substitute for human judgment or craftsmanship. He points to the tension inside his own product, where agentic workflows can generate outputs quickly, but a trained designer can still often make a better call, faster, than a system “burning tokens” hoping to land on the right answer. That belief extends to his outlook on enterprise AI adoption broadly — companies want sovereignty over their data, auditability of every decision an agent makes, and systems that reflect their own institutional tradecraft rather than generic, ungoverned tool sprawl.

Looking ahead, Modi doesn’t believe a single model will dominate the market. Instead, he expects a world where organizations select the best model for a given job, demand governance and repeatability, and increasingly value domain-specific expertise over general-purpose capability — a dynamic already playing out in industries like investment banking, where firms are keeping headcount flat rather than replacing their top revenue generators with AI. For founders building in this environment, his advice is to stay unusually close to the sales process, resist the urge to force early product bets past their expiration date, and treat networks and reputation — not aggressive sales tactics — as the real engine of enterprise growth.

Key Takeaways

• What looks like a “pivot” is often just a founder responding responsibly to new market information — don’t force a square peg into a round hole once the data changes
• Domain-specific expertise remains a durable advantage, even as general AI capability expands rapidly
• Enterprises are prioritizing data sovereignty, auditability, and repeatability over flashy tools that fragment into ungoverned silos
• The most effective enterprise sales motion is often the least “sales-like” — bringing smart, relevant people together and letting trust build organically
• CEOs of AI-enabled companies benefit from staying directly involved in sales conversations to keep product direction grounded in real customer signal
• Human craftsmanship still carries a premium in design and creative work, even as AI accelerates production of first drafts

Conclusion

Shaun Modi‘s journey from designing lunar vehicles for NASA to founding Capitol AI illustrates how deep product and design experience can translate into durable entrepreneurship, especially in a market as fast-moving as enterprise AI. His account of Capitol AI’s early pivot away from health insurance, its current partnership with EY, and its expansion into banking, private equity, and national security offers a grounded look at how innovation and disciplined investment strategies intersect once a startup finds product-market fit. For founders and investors evaluating where AI governance, model-agnostic infrastructure, and domain expertise are heading next, this conversation is a useful benchmark for what responsible, well-positioned entrepreneurship looks like in 2026.

Follow our show on your podcast channel of choice or listen below.