Gabe Stengel - Building Investing Superintelligence - [Invest Like the Best, EP.492]

Summary

• Stengel’s two-year frame: firm redesign, not just waiting for more capable models. He invokes Dario Amodei’s “country of geniuses in the data center” and asks what Goldman, Millennium, or Citadel would actually do if those geniuses showed up tomorrow — his answer is that “figuring out how to apply AI into the investment life cycle is the biggest challenge over the next five years for every great investor,” while Jane Street needed fifteen years to build its dominant franchise.
• Rogo’s product eras map directly onto model eras: o1 Pro made it a reliable search tool; Opus 4.5 at the end of 2025 and beginning of this year made it “capable of basically anything a junior investment professional or junior banker was doing.” Stengel is candid about the “first mover’s disadvantage” — early applied-AI products stank because the models weren’t there — but argues that if you were right about the end state, “when they get there, it’s magical.”
• Counterintuitively, Rogo’s core market is dealmakers, not public equities, because private-market coordination and transaction plumbing remain heavily human. Public markets have more available data and existing automation, while much of Rogo’s surface area is “the underneath of the iceberg” — data rooms, DDQs, CRMs, and LP reporting. The generalizable lesson for vertical AI: build where much of the last-mile wiring is missing and “perpendicular to what they [the labs] want to build” — Anthropic won’t build a data-room business; for them that’s “stopping on the side of the road to pick up a penny” en route from $100B to $1T in revenue.
• The economics of investing exceptionally justify extreme token spend: “investors are happy to pay fifty thousand dollars for one really good idea.” The bleeding edge Stengel imagines is ten thousand agents per PM “fraternizing, talking about ideas, pontificating” for 24 hours to surface a single idea — and he insists the models are already “smarter than anyone I know”; what’s missing “is plumbing” plus unsolved compaction/memory for agents working one-to-many across a firm.
• Rogo prices per seat like Bloomberg or FactSet, but Stengel wants to skip usage-based pricing entirely and jump to outcomes: “what if I just charge you for every good investment idea I give you?” He can tolerate customer-cost misalignment today because “we are one percent of the way into our product roadmap” — the vision is a Bloomberg-like exchange where agents, not humans, negotiate and transact, so “it’ll take five minutes for KKR to figure out, can I sell this portfolio company to another sponsor, not five months.”
• For incumbent firms, this is a major innovator’s dilemma in the category, and his diagnostic question is brutal: if 90% of your enterprise value sits in people today but will sit in software, data, and systems in ten years, “what would you start doing?” His answer: extract “what lives in the latent minds of your best people” into systems you own — prescient exemplars, including a co-founder of a firm he names as “Molus,” John Momtazi, sounded “crazy batshit two years ago” wanting digital clones of their best bankers.
• The founding story is a lesson in survivorship: 40 investors passed on the Series A — “like being broken up with by 40 girlfriends” — before Keith Rabois came in. San Francisco investors underappreciated finance TAM because they lacked intuition for businesses such as ION Group, Bloomberg, S&P, FactSet, PitchBook, and Morningstar; the product “was wrong half the time”; and nobody had watched him chew glass. Pat Grady’s current board logic: if everyone in finance makes an AI buying decision in the next 18 months, the only question is “do you think this plan is aggressive enough or no?” — and the answer was no.

Deep dive

1. The next two years are about reinventing firms, not waiting on models

• Patrick opens with the uncomfortable observation that Rogo is “eating a lot of the core functions” of analysts and even PMs. Stengel’s response reframes it: two years is easier to reason about than twenty, and in that window “the best investors are going to be figuring out how to reinvent their own firms and reinvent themselves.”
• His anchor analogy: Jane Street took fifteen years to build the dominant quant/market-making franchise, and today’s best investors will spend two to five years integrating AI. Borrowing Dario’s line about “a country full of geniuses in the data center,” he asks what Goldman, Millennium, or Citadel would do if those geniuses showed up — “it would probably take them a while to figure out how to change the way they work.”
• On which human skills survive: he prefaces with humility — “I worked in finance for two years, and so I am a student of these guys” — and says the jury is still out on whether AI replaces judgment. But if a Move 37 moment, when Lisa Doll saw something he could never see, starts to happen in public equities, the durable skill becomes gathering “data and inputs into their model that no one else will have”: field time, expert networks, relationship graphs.

2. Product eras tracked model eras — and the first-mover’s disadvantage was real

• Stengel tried starting Rogo twice before it worked: a high-school app tracking the equity exchange ratio of two public companies as they were merging (“terrible”) and a college paper commercializing AI for financial econometrics pre-GPT-3 (“nothing worked at all”). The real business launched with GPT-3, pre-ChatGPT, when everything was demoable and nothing worked.
• The functional eras: o1 Pro was “the first time you got enough reliability” for tasks like calculating a metric across twelve quarters without being so annoying you’d do it yourself; then with Opus 4.5 at the end of 2025 and beginning of this year, models became “capable of basically anything a junior investment professional or junior banker was doing, as long as you gave it the right instructions and context.”
• His honest framing of the trap: “there was a first mover’s disadvantage for a lot of applied AI companies” — you knew where the world was going, users tried the early product and said “this is terrible, this is garbage.” But if you were right about the end state, “when they get there, it’s magical.”
• The craft that converts capability into adoption: an MD who “has not logged into a computer in ten years but does have an iPad” emails a deck markup to Rogo’s AI analyst and gets it back in twenty minutes instead of two days, with full auditability surfaced to the junior on the deal — details you only know “if you have the kind of spidey sense for what the job is.”

3. The models are already smarter than anyone — the frontier is plumbing

• The bleeding edge he’s building toward: imagine every hedge-fund PM with “ten thousand agents that were just kind of fraternizing, talking about ideas, reading through the notes, pontificating,” delivering one idea after 24 hours of debate. The economics work because “investors are happy to pay fifty thousand dollars for one really good idea” — few domains justify that token burn for a single insight.
• His flat assessment: “the models are smarter than anyone I know, anyone I spend time with, and it’s plumbing” — connecting context, thesis, and workflow is what’s actually cutting-edge.
• Despite public equities looking like the obvious application (“all the data’s available”), Rogo’s ICP is dealmakers — people buying, selling, and coordinating transactions: preparing and unpacking data rooms, answering DDQ questions on customer concentration, mapping deals to a firm’s investment philosophy. Half the product is “the underneath of the iceberg”: updating CRMs, portfolio monitoring, and LP distribution.
• On expanding to public markets, a rare admission of ignorance: “I don’t know what it takes. I’ve never done it.” He gets board pushback to broaden the ICP but sees so much depth in dealmakers that “I can build a huge, huge business just concentrating where I am today.” Patrick’s distillation, which Stengel endorses: the vertical-AI opportunity lives where “there’s lots of plumbing that’s not yet built” — private markets are attractive precisely because coordination, standardization, diligence, and transacting are still largely done by humans.

4. Counter-positioning against OpenAI and Anthropic: build perpendicular, own the harness

• Architecture evolved from “this Rube Goldberg contraption” of sixty model calls toward giving a smart model the simplest, highest-quality tools — plus compliance lineage built so that “if someday some Delaware court judge makes AI inputs and research discoverable, you have actually done it all the right way.”
• The lab-competition answer: “you want to build things that are perpendicular to what they want to build.” Finance is “a collection of different niches with different data sets, different definitions of good, different regulatory requirements,” and Rogo can reach $5B in revenue by going deep — for Anthropic that would be “like stopping on the side of the road to pick up a penny” on the path from $100B to $1T. Concrete examples: pixel-perfect MNPI audit plumbing, and a compliant, agent-integrated data room — “I don’t think OpenAI or Anthropic will ever want to build a data room business.”
• Patrick’s synthesis — of data, model, and harness, harness matters most — draws Stengel’s best riff: Claude Code and Claude Cowork ran up usage on similar models because the harness “allowed the models to exercise more of their long-running capabilities.” Human intelligence itself is “all these different microservices in your brain” — emotions as triggers, a great investor’s judgment as “spidey sense” that fires recall into a creative decision.

5. The playbook for investing in Rogo-like vertical AI businesses

• Three filters if forced to be an investor in the category: (1) enough complexity in data, systems of record, and deployment models that solving them is a wedge — if anyone can walk in and sell basic ChatGPT, you never earn time to build the perpendicular stuff; (2) domain expertise — Rogo has over a hundred people from investment banks and firms, and “our job is really to catch the change in the models”; (3) willingness to “constantly reinvent the core product and constantly willing to slash it to nothing” — anyone wedded to a terminal or fixed UX “is not going to be agile enough” for six-month step changes.
• The reinvention model he admires: Max Levchin’s Affirm rebuilds its fundamental ledger every year — it retains engineers and stops the system ossifying. Rogo does the same with its harness: “it would be impossible if we were at a local minimum because the models are changing so quickly.”
• The capability gap that would change everything: compaction — an agent maintaining coherent memory across a hundred conversations. Almost all agents today are one-to-one; once they work across a hundred colleagues in Slack, “the compaction problem just scales exponentially.”

6. Per-seat today, outcome-based tomorrow — and the exchange endgame

• Rogo is “a classic enterprise software business,” priced per seat because buyers bucket it with Bloomberg, FactSet, Capital IQ, and PitchBook — every deal needs an AE, solutions architect, and handshakes. Contrast: Anthropic selling to Rogo was “very easy, and the amount that we pay them has risen exponentially without a human in the loop.” The consequence: Rogo must build a go-to-market machine “five times faster than most enterprise sales organizations ever have to.”
• On pricing evolution, he wants to skip usage-based entirely: “what if I just charge you for every good investment idea I give you?” Token-based pricing invites disputes over value; with a good idea, you can see how much money you earned. On the misalignment of heavy usage under seat pricing: “we are one percent of the way into our product roadmap. Ninety-nine percent of the innovation for capital markets is in front of us.”
• The end-state vision borrows Bloomberg’s strategy — data in the door, workflows on top, then the exchange — except “I need to build the communication channel for the agents to transact.” The mortgage analogy carries it: nobody thought you’d remove the human from home loans; now 40–50% of mortgages are delivered online. Predictions: raising capital as easy as buying an equity on Robinhood, KKR testing a portfolio-company sale in “five minutes… not five months,” assets priced in an order of magnitude less time, markets “more transparent, more liquid, more efficient.”

7. Banks first, and the individual-to-firm productivity gap

• The customer base is mostly large banks — partly biography (he did buy-side M&A coverage), partly structure: banks are “the distribution channel for the rest of finance” and have the most seats — landing Bank of America reaches far more people than ten single-portfolio-manager public-equities investors.
• On enterprise adoption depth: individual productivity is exploding while firm-level productivity is unmeasured. An MD tells him, “I sent five pages to a client that before I would have had to go back and forth with an analyst on over three days to create, and I made it in ten minutes myself” — bankers who haven’t opened Excel in twenty years. The open question is strategy: cut costs, or enter new markets? JPMorgan’s announced SMB M&A push illustrates the opportunity: if you have a banker who can be a deal team of one, you can enter markets that previously made no economic sense.
• On hallucinations — a word Patrick notes has vanished from conversation — Stengel’s reframe: “it’s actually more important to be auditable than it is to be accurate.” An answer whose assumptions and sources are visible is actionable even when wrong; and as copilots become autopilots making decisions, you’ll need to debug agents like bad investors — “was there an incorrect data input? Did someone lie to them?” — especially for regulators.
• Internal enablement runs on a company brain called Shrek: every internal conversation recorded and piped in, proactive and reactive — it can prep you for Thursday’s private-credit meeting with use cases and ROI metrics before you ask.

8. Forty rejections, Rabois’s check, and Pat Grady’s aggression test

• The Series A low: David Tisch introduced him to 40 top investors — Sequoia, Kleiner, Benchmark, Thrive — and every one passed after dinners, ICs, and weekends. “It’s like being broken up with by 40 girlfriends.” Keith Rabois arrived a month later saying “it’s basically just Harvey for finance. Why would I do it?” — and Stengel’s comeback: “if it’s not contrarian, why did every single one of your friends just say it was a bad idea?”
• Why they passed: San Francisco underappreciated finance TAM because it lacked intuition for businesses such as ION Group, Bloomberg, S&P, FactSet, PitchBook, and Morningstar; the product “was wrong half the time” (“guys, you know we’re on the exponential”); and nobody had enough data points on him. The same investors led later rounds after watching him deliver on every disbelieved claim.
• The aggression lesson, via Pat Grady at last Wednesday’s board meeting: if everyone in finance makes an AI buying decision in 18 months and will buy something regardless, “the only thing that matters is that you can blitz the market as quickly as possible… do you think this plan is aggressive enough or no?” It wasn’t — “because I was being soft.” His venture math: “it’s fine if it increases the likelihood that I fail by thirty percent if the odds that I become a hundred-billion-dollar company also increase by twenty percent.”
• The texture of building: “chewing glass” — lost recruits, products washed over by the next model release, 40 nos — and an emotional confession: “I’m a deeply paranoid, deeply insecure, deeply scared person… every day it feels like you’re on the knife’s edge of a thousand things collapsing.” The paranoia is functional: business building is “a game of compounding momentum.”
• His advice to young founders is reps and relationships: be direct and transparent about what worries you and where you want to go, while believing a small-probability outcome is possible without pretending it is certain.

9. Incumbents face a major innovator’s dilemma — and some alpha may stay human

• Finance hasn’t had a disruption shock in a long time — fund momentum makes private-markets businesses hard to kill — but now “hundreds of AI native disruptors” can attack. His question for incumbents: if 90% of enterprise value moves from people to software, data, and systems within ten years, start “taking all of what lives in the latent minds of your best people and putting it into a system that you own.” Then audit every step of the deal lifecycle: do you have a relationship or context no one else has, “or do I just think I’m smarter and better read… in which case, AI is gonna obviate that.”
• The exemplars who look prescient today “sounded crazy batshit two years ago” — including a co-founder of the firm he names as “Molus,” John Momtazi, who wanted digital clones of the best bankers so every junior could leverage their context and relationships to generate revenue. Inside Rogo, AI-nativeness is enforced with humor: a monthly stack-ranking of AI-tool usage per division, with the bottom user’s printout posted “with a dunce cap.”
• His stated uncertainties, hedged as spoken: how quickly private markets standardize (“AI should fix that,” but a regulatory or market force accelerating standardization is “a little bit out of my control”), and “how much alpha will be left in the human relationships” — a twenty-year small-business owner may still want a handshake, but “for a lot of sponsor-owned businesses or things like secondaries or private credit or GP/LP secondaries, I do think it can be fully automated.”
• The closing warning to fellow builders: winners “are gonna become black holes for talent and capital and brand”; “everyone is getting pulled into the big leagues and is having their welcome to the NFL moment” — fail to become that black hole and “you are far more at risk of being roadkill of a lab or a company that can.”