Gabe Stengel - Building Investing Superintelligence

Invest Like the Best with Patrick O'Shaughnessy

Gabe Stengel, CEO of Rogo, discusses the evolution of AI-driven tools for the finance industry and the necessity of building specialized infra

Key takeaways

• The true value in vertical AI lies in the 'last mile'—handling compliance, regulatory workflows, and auditability rather than just raw intelligence.
• Applied AI companies face a 'first movers disadvantage' when building for models that are not yet capable enough to support the intended product vision.

Main topics

• Vertical AI in finance
• Evolution of LLM capabilities

Notable quotes

The best investors are going to be figuring out how to reinvent their own firms and reinvent themselves.

Conclusion

The episode highlights that the next frontier of AI value is found in

Transcript preview

Speaker 3 (0:00) Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5 % annually on average so you can stay focused on growth. Ramp just opened for business in the United Kingdom. So if you're running a business in the UK, you can now use Ramp's AI-powered finance platform to manage cards, expenses, bills, approvals, and accounting all in one place. I run my business on Ramp, and so should you. Learn more at ramp.com slash invest. Open AI, Cursor, Anthropic, Perplexity, and Vercel all have something in common. They all use WorkOS. And here's why. To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO, SCIM, RBAC, and audit logs. That's where WorkOS comes in. Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOS APIs to gain all of them on day zero. That's why so many of the top AI teams you hear about already run on WorkOS. WorkOS is the fastest way to become enterprise ready and stay focused on what matters most, your product. Visit WorkOS.com to get started. Speaker 3 (1:13) You and I have talked many times about this basic question that I'll start with over the last couple of years. You are effectively trying to build investing superintelligence tools to help investors do their job much faster, better, cheaper, easier, higher quality. But it's starting to feel like, wow, we're really eating a lot of the core functions that even a very smart analyst or even portfolio manager was doing a couple of years ago. How do you think about that trajectory as you've seen it and lived it so far and where it's going over the next two years? Speaker 2 (1:42) I think two years is actually easier to reason about than 10 years or 20 years, because in two years, the best investors are going to be figuring out how to reinvent their own firms and reinvent themselves. And if you look at what happened to market making and quant trading, Jaden Street took 15 years to build the dominant franchise. And the world's best investors today are going to spend the next two to five years figuring out how to integrate AI into what they do. And Dario has the great line about everyone's going to have a. data center full of geniuses or a country full of geniuses in the data center, what would Goldman do? What would Millennium do? What would Citadel do if they had a country full of geniuses show up? It would probably take them a while to figure out how to change the way they work, how to take advantage of that, how to integrate it into their system. I think figuring out how to apply AI into the investment lifecycle is the biggest challenge over the next five years for every great investor. There's been Speaker 3 (2:32) many companies on this trajectory where the product was Cognition is very famous for like literally their ads now say, remember Devin, like it's good now. So lots of now clearly great companies with great products had a stage of an AI business where the product stunk and now it's excellent. If you think about the couple increases in capability that we've seen just from the raw models, could you do the same thing for the eras of Rogo and its product? You pick how many eras it is. I don't know how you frame it up, but like what it could do at each level up Speaker 2 (3:07) to and including today. Yeah, I actually tried starting Rogo two times before we got started. So in high school, I had a friend whose dad was an investment banker who wanted an app for trying to track the basically equity exchange rate of two public companies as they were merging. And so we tried using really old AI techniques to do that. Terrible. And then in college, before GB3 came out. We published a paper on AI assistance for econometrics and financial econometrics. And we tried commercializing at the time and nothing worked at all. And then when we actually started the business, it was when GBT3 came out, pre-ChatGPT. And so all the early days of Rogo, it was clear how kind of magical it was. You could demo things that were cool. Nothing worked at all. I mean, I would say since things actually started to work, the eras are very tied to the model eras, right? It was 01 Pro and then probably Opus 4.5. O1 Pro is the first time you got enough reliability where it was a good search tool at the very least. You could say, help me calculate this financial metric for this business over the last 12 quarters, and it could do it reliably enough where it wasn't so annoying that you would just do it yourself. And then with Opus 4.5 and the last year, the end of 2025, end of the beginning of this year. The models just 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. I mean, I think there was a first movers disadvantage for a lot of applied AI companies because you thought you knew where the world was going and you wanted to build a product for it, but the models weren't quite there. And so people would try it and go, this is terrible. This is garbage. You have a disadvantage. For us, we saw that too. But now what we've seen is. Well, if you were right about the end state and where the models were going and you were building towards that, when they get there, it's magical. And for me, a great product is all the feedback we get every day. People all day are saying, hey, this is transforming the way I work. I'm saving hundreds of hours a month. I am doing things I never could have done before. And so I'm smarter as a result and I'm able to make better decisions. And it's delightful. And I like using it. And, you know, it brings me joy in my day to day because the UX and the attention to detail and the craftsmanship is so obviously built for me and who I am. And so that's been the best part of the product. And you would attribute that to Speaker 3 (5:24) you took seriously all the compliance, regulatory workflow. last mile hookup stuff, and then the models became good enough. And all of a sudden, that was super valuable. Well, Speaker 2 (5:35) there's also small details of understanding how someone within one of these firms works and building for it. I'll give you an example. We make it so easy for a managing director at Speaker 3 (5:44) a Speaker 2 (5:44) bank to email a markup of a deck, which is how they're typically doing these workloads anyway, except sending it to an analyst and sending it to our AI analyst over email and then returning that markup in 20 minutes as opposed to. two days, and at the same time, alert the junior analyst on the deal, what's happening, and show them the full auditability of all the little markups that were made in case they want to weigh in. And that whole UX, that whole flow just makes it so much easier for this financial professional who is not logged into a computer in 10 years, but does have an iPad where they know how to mark these things up to actually adopt and use AI. And there are these small details of how you build a product that's great for a specific end user that you only know if you have the kind of spidey sense for what the job is. Speaker 3 (6:29) What Speaker 2 (6:29) is Speaker 3 (6:29) like the bleeding edge of what it can do that impresses you the most? What Speaker 2 (6:35) kinds of jobs? The coolest things that we're working on is taking these innovations like Malt Book. Imagine if every PM at a hedge fund had 10,000 agents that were just kind of fraternizing, talking about ideas, reading through the notes, pontificating, and then at the end of... 24 hours of debate just gave you one idea. And the reason you're able to do that is because investors are happy to pay $50,000 for one really good idea, whereas there's very few other domains where you can expand that many tokens just for one simple insight. What we're seeing right now, though, is that the models are smarter than anyone I know, anyone I spend time with, and it's plumbing. Connect it to your context, inform it about your thesis, tell it the way that you work, and try and integrate it into what you do. And so building out all the plumbing to actually collect that data, collect that context is what is cutting edge to me. It's always interesting to me for a product like this that Speaker 3 (7:24) you're opinionated about what it should be used to do. But in some sense, people can be creative with how they use it. So you get to sort of revealed how people want to use it. If I adopted a God's eye view of it's Monday morning here in New York City, lots of roger users are probably fired up and using it right now. If I could somehow see into every instance of the product being used. What would I see? Who are the people? What are the predominant use cases? How varied are they? Give us a sense of how it's being used right now as we record. Speaker 2 (7:51) Even though in so many ways, I think public equities is the best application of AI because all the data is available. And so it's just about being as smart as possible. That's a very brute force framing. Our early users in ICP and kind of core market is actually what I would describe as deal makers or people that are transacting, who are buying companies, selling companies. helping coordinate transactions. And so a lot of what we do is both make people smarter, but actually do the deal making. How do you prepare a data room? How do you unpack a data room? How do you coordinate the call with the third parties to discuss the data room? How do you go through all of the initial steps through closing of a deal? And so if you took a bird's eye view of all the folks using Rogo, I mean, it's people who are either on the sell side of a transaction or the buying side of a transaction and are using it to basically. prepare all the thoughts and materials to help execute that full deal. Whether it's putting things into a data room, this is the company's model. This is the PowerPoint that describes their customers. These are the answers to the DDQ questions on what customer concentration is. Or it's all the agents on the other side of that that are tearing through the data and mapping it to the firm's investment philosophy to say, oh, great, you know, is it lower than the concentration risk profile that we would want for this fund too? And then the components of the system. are people access it via all the classic channels you would access an AI tool, email, chatbots, proactive alerts, and those kind of things. But then Rogo is actually in a lot of the behind the scenes systems of these firms. Because when you're working on a deal, it's not just important for the human beings working on it, but you need to update your CRM. You need to update your portfolio monitoring system. You need to update the way that you distribute. information to your LPs after the fact. And so half of our surface area is actually the underneath of the iceberg of interacting with these different systems of record based on what the humans are doing over the course of the deal. Speaker 3 (9:48) So dealmakers today, when do you think you'll be able to give the same answer for junior analysts at a public equity hedge fund or something like this? There's a process to their workflow as well, but it's very, very different. You're right that it's interesting that my first intuition would be public markets are the best place to do this because there's so much data available. When do you think that Speaker 2 (10:09) transition happens? I think that for our business, we need to have all the requisite domain knowledge of what it takes to be a great public markets investor. I don't know what it takes. I've never done it. I haven't spent nearly as much time as I should have with the folks that are great at it. And we need to both hire out that domain expertise and then figure out. based on it, how to apply the systems that we have built to that market. I have extreme conviction in the fact that the underlying systems and tools and infrastructure we have built will be invaluable to that market. But now we need the great chef who can figure out how to piece it together and create that kind of end state product and that last mile delivery for public equities investors. I get pushed a lot by our board to think about expanding the ICP beyond just core banking. But the reality has been is there's been so much depth and tan in this dealmaker's vertical. And then for my end state vision of actually being the full infrastructure for private markets where people can transact very effectively, that is far more important to the dealmakers. Whereas for public equities, all that infrastructure, all those exchanges already exist. And so I'd like to serve them because I want to serve the most sophisticated, smartest users who have inordinate amounts of knowledge on the companies they track and the industries. they follow. And I'd like to make them even smarter because that sounds super cool. But I can build a huge, huge business just concentrating where I am today. Speaker 3 (11:37) Interesting. So one takeaway from that would be a lot of the opportunity to build an AI business in a vertical is somewhere where there's lots of plumbing that's not yet built. Yes. And then applying Speaker 2 (11:47) your down on top of that. Exactly. I mean, part of the reason private markets are so attractive is because it's. all done by humans, the coordination, the standardization, looking into things and the actual transacting. Whereas public equities, a lot of it has been automated. Based on what you know, what skills Speaker 3 (12:04) do you think investment professionals, broadly speaking, public and private, should think about being or becoming more valuable as time progresses and which skills become, it's kind of obvious the skills will become less valuable. What the hell are we going to do? I think can do the soup to nuts, diligence and... I see memo and objection handling and all this kind of stuff. Like that's a big part of a job for at least a junior person in the investing world. So what skills do you think people will still matter a lot Speaker 2 (12:32) in a couple of years? I mean, I want to preface it all with, I worked in finance for two years. And so I am a student of these guys, just as much as I'm fascinated by the technology and want to figure out how to use it. But the world's best investors have a way of. figuring out what matters and exercising their own judgment across a range of topics. And I think jury's still out on whether or not that is something that AI can eventually replace. When Move 37 happened and Lisa Dahl saw something I could never see, if that starts to happen in public equities, yeah, it's going to really change what matters. And if that really changes all of how that works, I mean, I think the core skill set is folks who can go out and gather data. and inputs into their model that no one else will have. If you can spend time in the field, if you can speak to experts, if you can develop a relationship graph of folks who can inform your model, maybe you're not the one that needs to calculate what your move 37 would be for a great public equities investment, but you can actually feed your model with data that no one else has too. Speaker 3 (13:31) Vanta automates compliance so your team can spend less time on security reviews and more time getting customers. Vanta cuts audit prep by 82 % and gives you instant up-to-date proof of trust without all the manual work, so you can close deals faster and with less friction. Customers report a 526 % return on investment, and more than 16,000 companies use Vanta, including Ramp, Harvey, and Snowflake. Get started with $1,000 off at vanta.com slash invest. Ridgeline is the first end-to-end system of record with embedded AI for investment management firms running portfolio accounting, reconciliation, reporting, trading, and compliance on one unified platform. Firms are moving off legacy technology and onto Ridgeline because of how far ahead Ridgeline's AI features are compared to anything else in investment management software. I've been hearing from a lot of investment managers about AI, and they fall roughly into two camps, with some unsure where to even start and others convinced they can build their own order management system over a weekend. The reality is that running an investment firm will always require governance, controls, and a single source of truth for your data. And no amount of AI enthusiasm changes that requirement. Ridgeline is built on exactly that foundation, which is why I believe that the firms that come out ahead in the AI era will be the ones running on Ridgeline's unified platform. If you're serious about your firm's AI strategy, Ridgeline should be part of that conversation. And you can request a demo at ridgeline.ai. Speaker 2 (14:59) Maybe it's a good time to talk about, I think about the assembly Speaker 3 (15:02) line. Like if Rogo puts out some really useful output, I want to learn about each component part of the system that leads to that output. And I'm most especially interested in the data that you yourself have and use and buy and build and whatever, and how you do that, how you think about data and then how you think about model. And these are both questions, one that I'm curious about your specific business for, but I also think that there will be some business like Rogo built in basically every vertical. And so I'm curious. what can be abstracted to other professional services or other verticals that are interesting as well that will fall to AI progress. So talk us through specifically like the data and model piece and how that's evolved over time. Speaker 2 (15:39) The early days of Rogo was kind of this Rube Goldberg contraption where like you had 60 different model calls. A question comes in, you try to say, what companies is Patrick asking about? Now, what are their tickers? Now, how do I feed those tickers into an API call to Bloomberg or Faxit or some internal data set? Now it comes back and I need to call a different model to pull it all together. And as the models get smarter, you kind of want to be less prescriptive, less of a Rube Goldberg machine. And just think about what are the simplest, best, highest quality tools. And the same way that if you have the world's smartest human being starting here tomorrow, trying to be a great banker and investor, what are the tools that they would need that are not just intrinsic to being smart, right? What are the data tools? What are the ways of going out and gathering information? What are the ways of auditing its own work? And