A clunky chatbot, a single Slack ping, and a $2 billion idea
Gabriel Stengel runs Rogo, the AI platform that more than 250 financial institutions now point at the work nobody at the firm ever wanted to do. Earnings summaries. Benchmarking net revenue retention by hand. The pitch deck due at dawn. Rogo's autonomous agent, named Felix, screens deals, drafts CIMs, runs data-room diligence and chases buyer outreach - the synthesis grind that used to eat a junior banker's night.
In April 2026 the company raised a $160 million Series D led by Kleiner Perkins, with Sequoia, Thrive Capital, Khosla Ventures and J.P. Morgan all in. The round valued Rogo at roughly $2 billion. Four months earlier it had been worth about $750 million. Bloomberg summed up the whole arc in a single headline: junior bankers sick of grunt work built a $2 billion tool to do the job.
The origin is smaller and stranger than the valuation. At Princeton, Stengel and classmate John Willett built a chatbot for econometrics as their senior thesis. It was, by Stengel's own account, a crummy, slow thing. They graduated in 2020, scattered into finance - Stengel to J.P. Morgan, Willett to Lazard - and might have left it there. Then OpenAI released GPT-3 to developers that summer.
Stengel started playing with it and pinged his old thesis partner. "This could make that crummy, slow thing we built before kind of great," he told him. In January 2022 the two quit their finance jobs, pulled in Tumas Rackaitis as the third co-founder, and started building Rogo full-time: a kind of ChatGPT tuned to the way Wall Street analysts actually think.
For the first 24 months nobody wanted to talk to us.
That line is the part founders rarely frame. Two years of doors that stayed shut. Wall Street does not hand its most sensitive numbers to a startup on faith, and the early Rogo had no logos to lean on. The first paying customer did not arrive until late 2023. After that, the curve bent hard - a Series C in January 2026, a Series D by April, and a client roster that now reads Rothschild & Co, Jefferies, Moelis, Nomura and the very Lazard where Willett once pulled all-nighters.
Stengel is careful about the story finance tells itself when a tool like this shows up. His framing is blunt: Rogo replaces work, not people. He reaches for the ATM - the machine everyone assumed would gut the teller, which instead made branches cheap enough to multiply, and tellers more numerous than before. Free an M&A advisor from manual benchmarking, his argument goes, and the advisor does not vanish. The advisor moves up - into consulting, geopolitical read-outs, the judgment a spreadsheet can't fake. "The opportunity to expand your service and enrich it is now kind of endless," he has said.
His method for keeping pace with the technology is almost suspiciously plain. "The best way to stay apprised is just being very open-minded and non-dogmatic about what it can do." No grand theory of the singularity. Just a refusal to decide in advance what a model can't handle, then watching it handle more.
By late 2025, Forbes put Stengel and Willett on its 2026 30 Under 30 list in finance - young founders building the thing that automates the industry that minted them. The recognition is tidy. The underlying bet is bigger: that the most sophisticated firms in the world stop merely automating tasks and become, in Stengel's phrase, AI-native - with agents that work across the whole firm and "get smarter with every deal."
Six lines that explain the bet
This could make that crummy, slow thing we built before kind of great.
We're replacing work, we're not replacing people.
The best way to stay apprised is just being very open-minded and non-dogmatic about what it can do.
The institutions at the forefront are moving beyond automating tasks to becoming AI-native firms - agentic systems that get smarter with every deal.