Rogo, founded in late 2021 before ChatGPT, automates junior banker work like earnings analysis, benchmarking and pitch decks for banks and funds.
Gabe Stengel argues AI value is shifting to the application layer, since personalization and context make switching hard and no model provider stays six months ahead.
Stengel predicts AI will replace workflows rather than people, with banks partnering with startups, though he expects a mindset shift over five years.
Rogo bets finance's AI value lies in applications, not models
Rogo is building an AI analyst for Wall Street on the bet that AI's value accrues to the application layer. Licensed data, custom tools and workflow knowledge are hard to copy while frontier models converge. Stengel sees the same product eventually serving companies too small to afford a bank.
Takeaways
AI's value is shifting to the application layer. Rogo post trains frontier models instead of building its own. No provider stays more than six months ahead, Stengel says, and personalization and context make switching hard. OpenAI and Anthropic may eventually target finance, but he expects Rogo's workflow and evaluation lead to hold.
Rogo is deep research wired into finance. It connects to data sources such as FactSet and Capital IQ, plus a bank's internal data. Rogo builds its own tools, including Excel and a filing search, and trains specialized models to use each. Ten company profile slides take about thirty seconds instead of hours.
The product grew from Stengel's Lazard analyst work. A Princeton senior thesis chatbot for econometrics sparked the idea. At Lazard he built models and decks, and public information books, 200 page research packets that seniors want within a day. Rogo now helps investors research and diligence companies and form theses.
Early buyers treated AI as snake oil. Rogo was founded in late 2021, before ChatGPT, and had no real customers until late 2023. Product-market fit came after it added unstructured data and sharpened its pitch. Kevin Ryan, once CEO of DoubleClick, led the pre-seed round.
Security built early speeds enterprise adoption. Customers demanded multi-cloud and on-premise deployment and permissioning, so Rogo treated security as a core value proposition. Stengel says enterprise grade security lets it move faster into large financial institutions.
Finance sales stay top-down, not product-led. Strict information security policies make product-led growth hard, so Rogo sometimes starts with five seats and upsells. Stengel does not expect a Notion or Airtable style bottom-up spread. A typical deal involves about 40 stakeholders and is political.
AI replaces workflows, and banks will partner. Stengel says banks excel at relationships and judgment, not building AI products, so they will buy from startups. He expects a mindset shift within five years, before firms rethink business models, and points to Harvey and Offdeal.
What Stengel expects next
Within five to ten years, Rogo becomes the most effective analyst on Wall Street.
Rogo wants to offer M&A advice and fundraising directly to small private companies that cannot pay Goldman Sachs fees of $5m.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Stengel argues a company with a $20m enterprise value that skips a sell side adviser gets a price 20% worse.
He suggests AI could produce fairness opinions at a fraction of a banker's fee, though human bankers with relationships will still matter.