Invest Like the Best with Patrick O'Shaughnessy
00:23:03 · Patrick O'Shaughnessy, Gabe Stengel
[00:23:03] S3: What's an example of that that you've done like a tear down and rebuild? [00:23:07] S1: The story, the anecdote I'm most inspired by is Max Levchin who talks about how at a firm they rebuild the fundamental ledger technology every year. And they rebuild it for a few reasons. One is it's the most interesting engineering problem and so all the engineers wanna work on it. So it's a good way to retain talent to teach engineers about the core fundamental business of a firm. But But then number two, it's a good way to make sure that system doesn't ossify and it's constantly improving. And so we do that same exact thing for our harness and the core agentic system. We are constantly looking at it realizing we're not even at a local minima. It would be impossible if we were at a local minima because the models are changing so quickly and we need to redo the whole thing. [00:23:49] S3: What's something that the models currently cannot do that if they could would really change the nature of the product?
00:23:49 · Patrick O'Shaughnessy, Gabe Stengel
[00:23:49] S3: What's something that the models currently cannot do that if they could would really change the nature of the product? [00:23:55] S1: Compaction. So it's if you have a 100 conversations with a single agent, how does it make sure that it's actually remembering the right things and compacting its memory into an amount of tokens that it can use every time and it has enough coherence and context on who you are and what you care about to make it feel like it's a true person that you're speaking to that you know learns more and more about you. That's a hard problem and it compounds exponentially when you think about agents not just as a one to one. Right now almost all agents are one to one. You use ChatGPT individually. You use Copa individually. You use Gemini individually. As soon as these are actually things that consort with a lot of colleagues or in a Slack channel with a 100 people have to work across an entire company, now the compaction problem just scaled exponentially because it's having conversations with a 100 different people and needs to be able to coordinate across those things. And so being able to take all of that memory, those interactions and actually lodge it into the mental model or brain of that agent so that it can be persistent so that it can actually maintain context over the course of a bunch of interactions, that's something that the models are not great at today.
00:25:02 · Patrick O'Shaughnessy
[00:25:02] S3: What do you think the major kinds of AI software businesses there are? So we've got companies like a Cognition or a Cursor or something that can grow unbelievably quickly. And I'm especially curious for you to compare this like the old classification system for software companies. I'm curious how people buy ROGO, what kind of category you would put it in. Is it usage based? Is it seat based? Is it something else? Like, talk us through the how people wanna buy this stuff and what the emerging models are for AI software business.
00:25:30 · Gabe Stengel, Patrick O'Shaughnessy
[00:25:30] S1: We are a classic enterprise software business. We price per seat right now because our buyers are used to pricing per seat. They think of us in a similar category to Bloomberg, to FactSet, to Capital IQ, to PitchBook. And so we have to build a very human business. Every time you sign a deal, it requires an AE and a solutions architect or the sales engineer going in and shaking a lot of hands, explaining how it works, explaining how to integrate it. You can't sign one deal where the usage just rises a 100 fold. I look at how hard it was for Anthropic to sell to us very easy and the amount that we pay them has risen exponentially without a human loop because it's a token consumption model. There's a lot of industries where riding the coattails of token consumption isn't gonna work for enterprise sales and we're one of those. And so it's actually pretty interesting because we have to build a go to market machine five times faster than most enterprise sales organizations ever have to build. And so I do think there's the category that's just typical enterprise sales, but using AI models as a tailwind to build a 100 times better products. And then there are the token brokers, the token consumption businesses where you're selling in the parts of enterprise where they're used to buying usage based tools. Cursor, factory, cloud code and others. And so you can go much further commercially with fewer people. [00:26:45] S3: What's your prediction for how or if that will change in finance?
00:26:45 · Patrick O'Shaughnessy, Gabe Stengel
[00:26:45] S3: What's your prediction for how or if that will change in finance? [00:26:48] S1: I think that every business needs to go through two different pricing revolutions. You need to move to some sort of usage based and then you need to move to some sort of outcome based. For me, if I can figure out a way to skip the token base, skip the usage base, simplify it for my users and just wait until I can say, hey Patrick, what if I just charge you for every good investment idea I give you? Or what if I charge you for the quarterly report you send to LPs that I can do perfectly? Or what if I charge you for every sim that you create as a banker? I would much rather get there than have to figure out some random way of trying to assign dollars per token that is something that we're not gonna quite agree on because you're gonna spend a $100,000 on tokens and say well did I get a $100,000 of value and I don't really know. But you know what the value is to you of a good investment idea because you can actually see how much money did I earn. Or you know what the value to you is if you can produce a sim if you're a bank because you know what you charge these firms to actually sell the business. [00:27:47] S3: Presumably, you can't change your seat price on the fly dynamically at least with the same customer. So how do you deal with the problem of like, in some ways, misalignment with the customer for your business where if you do a great job and they use the thing way more, which costs you a lot of money, they become a worse customer?
00:27:47 · Patrick O'Shaughnessy, Gabe Stengel
[00:27:47] S3: Presumably, you can't change your seat price on the fly dynamically at least with the same customer. So how do you deal with the problem of like, in some ways, misalignment with the customer for your business where if you do a great job and they use the thing way more, which costs you a lot of money, they become a worse customer? [00:28:05] S1: The reality is is we are 1% of the way into our product roadmap. 99% of the innovation for capital markets is in front of us. And so what matters is that we're a good partner. We're a good steward of their AI strategy, and they wanna work with us in the future. [00:28:18] S3: It's so interesting to think about the shape of this in the future. You said you're 1% penetrated into the road map. Give us a sense of where you think this is all going from your product perspective, not industry wide. If you have that much of the road map ahead of you still, describe that to us.