Building at the Intersection of AI and Financial Services
Earlier this month, we co-hosted a panel in NYC with our friend Seth Rosenberg at Greylock, highlighting three exceptional founders building at the intersection of AI and financial services: Mitch Troyanovsky (Basis), Jamie Cuffe (Foundation), and John Willett (Rogo).
These are some of the most interesting applied AI companies in fintech right now, building autonomous agents for accountants, modern back offices for insurers, and financial reasoning engines for bankers.
The room was full of founders, operators, and investors, and the conversation covered product, GTM, pricing, talent, and what actually matters when building an AI-native company.
Here are five takeaways we’re still thinking about:
1. New York is the best place to build vertical AI in fintech.
If your users are in finance, insurance, or accounting, build where they are. That means NYC. As John shared:
“Our customers are here. We’re going to conferences, meeting CIOs, and working closely with them on product. You can’t do that from a distance.”
The founders also highlighted the talent advantage; people who came up through tech companies with NYC outposts and now want to go earlier-stage. If you’re building vertical AI in fintech, being local gives you an edge.
2. Go deep with early customers, your reputation depends on it.
This was a recurring theme across the panel: it’s better to go deep with the right partners than to scale too fast.
“We spent the first 3–6 months just building side-by-side with one customer,” said John. “There was so much foundational work to get right in terms of infrastructure, data, integrations, training.”
Early customers become your references, your feedback loop, and your reputation in the market. Revenue can come later, but trust has to be earned early.
3. It’s not about useful products, it’s about valuable companies.
AI makes it easier than ever to build something that works, but that doesn’t mean it lasts. Mitch summed it up well:
“There are lots of ways to deliver value that don’t create enterprise value.”
Long-term defensibility still comes from embedding deeply into workflows, accumulating context, and owning the last mile. The companies that win won’t just be useful, they’ll be essential.
4. Per-seat pricing doesn’t work when AI replaces seats.
Traditional SaaS pricing breaks down when the product itself reduces headcount. That’s especially true when you’re replacing outsourced workflows. Mitch explained how they think about pricing at Basis:
“You can’t give a CEO a per-token bill. We price based on a proxy for the work being done, like number of end clients.”
The key is to align with value delivered, not usage or logins. Whether that’s task volume, documents processed, or outcomes achieved, pricing should reflect what you’re replacing.
5. The best talent strategy is hiring engineers, not coders.
The most valuable hires today aren’t just writing code, they’re architecting systems, modeling workflows, and solving real customer problems. Jamie put it plainly:
“The best people in these companies are the ones who can span functions. They’re technical, but they also understand the product, the customer, and how to sell.”
In an AI-first world, taste, agency, and judgment are just as important as raw technical skill.
We’re grateful to Mitch, Jamie, and John for sharing their perspectives, and to Seth from Greylock for co-hosting.
There is so much opportunity at the intersection of fintech and AI, and we at BTV are incredibly excited to be investing in this category and city.
Want to join us for future BTV events, panels, and dinners? Sign up here, we’d love to see you.
