Gabe Stengel, CEO of Rogo, says the best investors will spend the next two to five years reinventing their firms around AI, comparing the effort to Jane Street's fifteen-year build.
Stengel says models became capable of basically anything a junior banker or investment professional did by late 2025, which turned Rogo's early 60-call system into a simpler product embedded in firm systems.
Stengel argues auditability matters more than accuracy for finance AI, since users and regulators must be able to trace and debug agent decisions, and he says Rogo's customers are mostly large banks like Bank of America.
Rogo's CEO bets AI will rebuild finance from inside firms
Rogo is betting that AI becomes the working layer of finance, and that a vertical company embedded in firms' systems will beat general chatbots. Models can already do junior banker work, so the scarce parts are integration, auditability and trust in an industry where distribution is a people problem.
Takeaways
Models now match junior bankers given context. Stengel says o1 pro made search reliable, and by the end of 2025 Opus 4.5 could do basically anything a junior banker or investment professional did, given the right instructions and context. The best investors will spend two to five years reinventing around it.
Integration, not the model, is the product. Rogo's early product chained about 60 model calls. It now favors simpler tools, a harness around the models and last mile integration. About half the product surface sits underneath, in firms' systems of record such as CRM and portfolio monitoring.
Labs will not displace vertical finance software. Stengel argues Anthropic and OpenAI will not build data room businesses or the deep workflow infrastructure finance needs. He sees finance as a set of niches where a company could reach $5bn in revenue, protected by data and system complexity.
Per-seat pricing fits finance buyers for now. Rogo sells per seat like Bloomberg, FactSet, Capital IQ and PitchBook, because buyers are used to it. Token consumption pricing, used by Cursor and Factory, will not work for enterprise finance sales. Stengel predicts every business will move to usage-based and then outcome-based pricing.
Auditability matters more than accuracy. Users must be able to trace and debug agent decisions, especially under regulators. Rogo's internal company brain, Shrek, is fed by recorded internal conversations, answers questions and prompts staff before meetings.
Deal making and fundraising will get much faster. Fundraising, asset pricing and deal execution will become much faster. Mortgages are 40 to 50 percent online, and KKR can decide on selling a portfolio company in minutes, not months. Rogo's plan mirrors Bloomberg's, from data to analytics to a layer for agents.
Adoption depends on firm-level productivity. Stengel says the open question is whether individual productivity becomes firm productivity. Rogo's customers are mostly large banks, which have the most seats and feed talent into later investing careers. Rogo publishes monthly rankings of AI tool use by employee.
The Series A that 40 investors rejected
David Tisch of Box Group, an early investor, introduced Stengel to 40 investors after Rogo had built for two and a half years.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
All 40 passed.
Stengel cites a weak early product, doubts about him, underestimated market size and San Francisco's lack of finance intuition.
Keith came around about a month after the rest had passed.
Rogo later acquired six financial AI startups and now has over 100 people with bank or investment firm backgrounds.