Episode Transcript
[00:00:00] Speaker A: Oftentimes I talk to people and they said their first experience with Claude was amazing. Their second experience with Claude was disappointing. Why? It's the second time around they didn't get the same answer that they got the first time around and they thought that they should get the same answer, but it turns out that it inferred something slightly differently, like, oh, wow, I have to think about this thing called determinism.
[00:00:20] Speaker B: Hey, and welcome to the Momentum Podcast, the show where fintech meets AI. I'm your host, Stan Altschiller, investor, founder and advisor to some of the world's top hedge fund funds, asset managers and data companies.
Each week we'll dive into practical AI strategies to help you cut through the noise, beat data overload, build smarter workflows, and turn your ideas into results.
We'll explore everything from AI tools for your daily workflows to building data driven growth engines for your organizations. Here you'll discover what's next in finance and how to stay ahead of the curve with AI.
Let's dive in.
Right? Hi, everybody and welcome to the Momentum Podcast. I'm your host, Stan Altschuler, and we have not one, but two awesome guests for you today. We have Eran and Omer from Omega Point. Welcome to the show, guys.
[00:01:16] Speaker C: Excited to be here.
[00:01:17] Speaker A: Yeah, thanks, Dan. Amazing.
[00:01:20] Speaker B: My pleasure. Omar, why don't you give us just a little bit of background, give us a little bit about yourself and about your journey and then Iran, why don't you jump in and do the same thing?
[00:01:28] Speaker A: Absolutely.
Well, thanks everyone for tuning in and really appreciate you having us on your podcast. Yeah, and it's. It's certainly been a journey. There is no entrepreneurial journey that isn't. One, mine in particular began at a place called Two Sigma Investments, which probably many of you guys have heard of. It was a much smaller company at that time, but very innovative and thoughtful in how it can use technology to really push the envelope on what gets done in financial services and investing and so on and so forth.
And one of the areas that was very, very interesting within Two Sigma was the concept of how do we actually merge or integrate the fundamental, the best of what fundamental investing can offer and the best of what quantitative investing can offer. How can we come up with really the nexus that makes it really shine and generate the maximum amount of return on investment for each idea that's generated? Also highlighted to me and other folks working on that, the power of being able to integrate technology into the broad investment management processes in the marketplace. And what was really interesting at the Time and we're, we're now talking a decade plus. When we started talking to a lot of folks in the marketplace, we realized how little systematic quantitative investing techniques were actually integrated besides some of the larger oriented, obviously places that you're all familiar with. So we saw an opportunity, right, where 99% of the market does not have access to, to the technology underlying data and analytics. Two Sigma has, and only few places in the market do. And this is really a democratization story, right? This is how things happen at Google and other places in the marketplace where you might be at a very large company. You're seeing where innovation is going. You realize that 99% of the market doesn't have that innovation today. And it's an opportunity to really bring that to the masses. So we've been on this journey at Omega Point to really bring the best of modern data analytics, large scale data analysis, obviously now AI and whatnot to the masses.
[00:03:30] Speaker B: Very cool. Yeah, no, thank you for that background. Iran, why don't you give us a little bit about your journey?
[00:03:35] Speaker C: Yeah. So I've been a technologist and an entrepreneur through and through. This is my fourth rodeo. And the theme across all of them has really been extracting signal from noise and making it just very actionable for the advertiser. That's what makes people love it, right? That's what makes clients love it. That's what makes frankly me excited about what we're building. Before this, I was working in kind of the social media management space. We were building products for marketers and that really helped brands connect with their audiences. If Samsung wanted to know, right, Are people talking about, are people talking about related to the Galaxy Phone? Someone needed to go to Twitter and figure out, are they talking about the phone, are they talking about the Cosmos, are they talking about the soccer team? And so that was all something that basically we did, right, in an automated fashion and made it actionable so that these marketers could engage with it. When I was talking to Omer, and he has obviously a tremendous amount, a wealth of experience in the investment management industry, realized that this is kind of the same problem, just in a different costume. If you think about what Omer saw, what he described it to Sigma, right, like there really, there was a lot of noise in the factors and the flows. So at Oracle, right, there was noise in the, in the tweets and the operator was a different Persona. But here there's just a lot of noise in what's whipping around the portfolio. And the operator is the pm. So excited to just Kind of bring those worlds together. And really we've been building and working and making software that people, some of the largest asset managers in the world love now for over a decade.
[00:05:17] Speaker B: Sounds like an amazing partnership and a lot of complementary skill sets kind of fusing together. A lot of fun working with your brother. I don't know, you're like, ready to kill each other by the end of the day. But I want to dig in a little bit into kind of what you said, Omer, about Two Sigma. So obviously, look, Two Sigma is one of the largest and most successful hedge funds in the world. No doubt about it. It's also known for its innovation, the level of sophistication. Right. They bring to investment management.
This is kind of like the. The optics of Two Sigma and very well respected all over the buy side by institutional investors, obviously, other funds in that seat, right where you had all those resources available, you were building actually those tools. You were, you were part of the team that was leading the charge and building those innovative tools.
And you mentioned this, but what clicked? When did you realize, like, wow, there needs, like, no one else has access to this. Did you talk to other managers where you kind of like, made that leap where you said, wow, there's so much room for improvement based on what I see here@tusig vs what everyone else is doing. Can you tell me that inception point where in your mind, you're like, I am going to take this risk and leave like the best hedge fund in the world, right? And my very, very comfortable job and awesome job, I'm gonna go out on my own and build this thing because it's that big a problem. Can you walk us through that?
[00:06:39] Speaker A: Well, when you pose it that way, no. It's pretty funny. I think that there was a book that came out a while back actually called when to Jump. It reminds me of this concept, right? It's actually Sheryl Sandberg's, I think it was one of her family members wrote this, and he obviously had this experience as well. It's like leaving a comfortable job and jumping into the unknown, the entrepreneurial kind of world. And I think that part of it has to do, I would say, with a couple of things that, that kind of click. Right. First of all, there, I do believe has to be this innate drive to want to create and build something in an entrepreneurial kind of a fashion and being comfortable with the risk. I don't think it's for everyone.
I don't think everyone in the world is just ready to jump from their job and do something. Some people like to do it as a hobby.
[00:07:23] Speaker B: Right.
[00:07:24] Speaker A: Which is perfectly fine, by the way. Probably safer that way. But to really take the risk, there needs to be, obviously it's a calculated risk. First and foremost. There needs to be some evidence that suggests that the market needs something like this. And I think that's what you're heading on. There was a lot of overwhelming evidence that I was seeing again and again, and the team at Two Sigma was seeing again and again and how much of a need there is one of the well known kind of products that I created with the broader team there called the Alpha Capture business. Alpha Capture product at Two Sigma had a firsthand, basically a receipt directly to these many, many firms right out there that we were working with and looking at the state of what their technology systems look like, what access to data that they had, what type of decisions that they were making. And it was clearly a gap, a very meaningful gap. Right. A gap that you can ram a truck through.
And so that was really the piece of it that was helpful. Right. We weren't in an ivory tower looking back and saying, he, hey, we're quants and we're doing all this cool stuff and I think everybody else needs this too, so why don't we try this out in the marketplace. We had access to over 3,000. It was about 3,500 investment professionals globally that were participating in this alpha capture program. So we actually had real conversations, real data, real understanding of where data was at. It was just an intersection of the internal drive and need along with lots of evidence that suggests that there's something
[00:08:48] Speaker B: that's a gaping hole and Iran for you, the journey. So you went from enterprise trade, enterprise software, to building kind of like risk infrastructure for the buy side.
Tell me about that. You did say that some of the problems rhymed. Can you dig into a little bit where it clicked like, oh my gosh, I can really add tons of value here, Omer.
[00:09:07] Speaker C: Yeah, absolutely. And so a lot of that came from. You have really similar kind of a signal to noise problem. Right.
There is so much data out there when it comes to the social media space and what a marketer or a salesperson or frankly any corporation has to do in order to engage with their audience. Right. If you look at the way that they're advertising or you look at the way that they need to connect from a just conversational standpoint, the way they need to be a part of every single type of conversation that ends up happening, that is very, that's kind of core to Their business.
Well, frankly, it's core to the PM's business, core to the investment team's business to have an understanding of all of the data that's relevant to the thesis that supports their, that supports their portfolio, that supports the investments that they're making. Yet at the same time there's just an overwhelming amount of this data. And trying to extract what's reasonable, trying to extract what is going to actually move the portfolio, that's a very challenging problem.
And so I had a lot of passion from trying to figure out how do you make that actionable?
It's no longer about trying to engage with somebody in a conversation.
Now it's about really figuring out what's whipping around that portfolio, figure out what is it, how can I get ahead of what's going to happen and make the appropriate investments at the right time. And so really for us, you need to make sure that you're providing the right insight at the right time, the one that's going to make the biggest bang for their buck so that they can then process it and take action in the right way for their portfolio.
[00:10:45] Speaker B: That's great. And if we think about, put ourselves in the shoes of your clients, right, the GPS, most of them are not like, they're not quant PhDs, right? They're not super, super technical. They're not like two sigma level of sophistication. There are, some of them are right, but it's all over the map. And if you look at the belly of the curve, I would say most of them are just really stock people, like fundamental investors. They read the filings, they understand the management, they understand everything around the moat. Those are the kind of folks, right, that make up the belly of the curve for equity and credit investors. They understand the business, first and foremost, credit worthiness and the potential for those folks.
And I know this question would be very, very different if I were to ask it 10 years ago than now, because now LPs are pushing them, right, to be. No, you have to know your factor. Like you can't just be, this is a good, this is a story stock. But, but for those folks, why don't you take us through? Why should, why should they be quantum mental? Why should they care about factors? Why should they look at Omega Point or the sort of risk management software that you provide, should they still do it? Or could it just be window dressing like say, tell the LPs yeah, we're factor aware.
Is that enough or not? Can you give me your view? Well, either one of you can answer that, sure.
[00:12:04] Speaker A: I'm happy to jump in. I think I have this discussion almost daily with fundamental PMs about it's not so much about should we care, but what should we care about. Maybe I think it's more on the nuance. Of course, there are. There's so much information that's coming at them, right. If you think about the morning when they wake up and have to be prepared for the market. Open, right. Remember talking to 1pm, he says, I have literally have to get through maybe several hundred sometimes. Well, given all the newsletters and things that come in through is maybe a thousand emails that I have to figure out what's relevant, what's not relevant in my inbox. Right. And there's everything about the companies that they cover, right. Individually, there is what's happening kind of macro land in the market. They have to synthesize all that information and decide how they want to be positioned ahead of the open. Right. That's most PMs kind of job, right? I mean, it depends on obviously the length of time, how they're thinking about it. They don't have to do this every day, but the idea is they have to be prepared, have to be prepared to understand what's important. And I think that there's enough evidence to suggest that the fundamentals that they are core focused on, right, that they understand the company and its peers and its industries have a meaningful impact on how they should be positioned. But that's not the only information that's required. And all it takes is probably once or twice for the portfolio to behave in a particular way where they scratch it. They say, dude, the fundamentals are fine. Nothing is going on. There's no new news, there's nothing happened. But it just. Stock just got hammered or the entire sector just got moved, moved 10% and now I'm underwater and my LPs are calling me and they're asking me what's
[00:13:41] Speaker B: going on, what to answer that.
[00:13:43] Speaker A: They first need to be able to answer that and say I'm comfortable with it. It's an overreaction. Here's what's happening, here's what the data says, I'm sticking to it. Or they need to say, all right, I need to get ahead of this because I'm not comfortable showing a drawdown of X percent to my LPs. I told them that I'm going to maintain my risk and my drawdown within a certain amount of band and if I'm exceeding that, they're not going to trust me anymore. So I need to make sure I prevent ahead of time the likelihood of these drawdowns coming from events that I'm not aware of.
[00:14:13] Speaker C: Right.
[00:14:14] Speaker A: And that is where I think this quantity factor, the whole sort of, you want to boil it all into a soup, just give the soup in a very simple way to somebody and just say, you don't need to focus on every little thing. You just need to understand what are the kind of non fundamental drivers telling you and what you should pay attention to. And I think that that is essentially what we're doing is distilling, as Ron says, signal from noise, distilling. What are the most important drivers you should care about that are maybe outside of your core deep fundamental knowledge of the company that will move the stock and create risk on your portfolio and
[00:14:47] Speaker B: sometimes it will move a whole family of stocks. This is the part that I realized that fundamental managers sometimes don't understand that a lot of their themes are strong investment themes, but they're also connected on a technical level. They could be part of the same shock or sell off that other investors hold, like crowding. They could be a factor rotation. There could be a bunch of other stuff like just like you said, it's not going to come up on the radar in terms of doing fundamental analysis and cash flow analysis and DCF and all that great stuff that you do. Very valuable. But it's going to hurt you in some other way. You just need to be aware.
Could you give me an example like maybe 20, 21 or something that's kind of recent AI bubble talk. Can you give me something that's like you guys are underappreciating this, what's going on and the undercurrents of the market and by translation in your book.
[00:15:38] Speaker A: Yeah, I think let's make it relevant today. We can, we can go back to 23, 22, 21 every year. There's an example of that happening somewhere. Right. I mean, the markets are so complex and there's so many different facets of it that we're seeing these types of characteristics. It comes down to the fact that at a simple level, to explain this, that your company that you're looking at from a fundamental perspective has correlations to other companies.
[00:16:02] Speaker B: Yeah.
[00:16:03] Speaker A: Right. Now, some of them are for reasons that you understand. Right. There's a particular sea change in the industry that you are tracking and managing. But there's also other correlations and other reasons that could be completely either outside or tangential to what you're looking at. And you need to understand those correlations. And you mentioned Crowding. Crowding happens obviously for several reasons. One is a lot of fundamental investors or investors really, really like, whatever, Tesla right now because they think that Elon's finally going to hit home run with the humanoid robot and whatnot, or Apple is going to release the latest and greatest iPhone that's going to blow everything out of the water. So it's going to be improvement in sales. So they're making a fundamental bet on the company. That's an idiosyncratic crowding. Okay, but then there's something called thematic crowding. Wait, the company is correlated to all these other companies that people believe are going to get these data center contracts. They may have not even announced that they have a data center contract yet.
[00:16:57] Speaker B: Right.
[00:16:57] Speaker A: But they, everybody assumes because of the nature of what's going on and the fact that they're part of that industry, Rising tide floats all boats. They're going to be part of it. And all of a sudden the company that you might be short, let's say, because you're like, oh, the business model isn't working, is suddenly up 20%. You're like, what is going on? 20 to 25% up for no reason. Oh, well, it's up because the theme, the data center supplier theme is up. That's something going on in.
[00:17:21] Speaker B: And it's indirect. It's indirect.
But the only way you can understand that is you can understand those indirect relationships. But like the whole theme is melting up.
[00:17:30] Speaker A: Exactly. And data is important. Yeah.
[00:17:34] Speaker B: And since we've pivoted to the data eran, a question for you. Is there like, what should managers think about? Is there some data that they must have or be subscribed to or be very disciplined about how they store so that they can use your system and answer these kind of questions?
[00:17:51] Speaker C: There's a lot of data right now that is widely available that people can go to the same vendors and acquire that data. They basically have that alt data, they bring it. It's kind of internal and it's supported within their investment process. But I think where this gets more interesting is that right now everyone's applying AI to that data. And as they're applying AI to that data, there is really a host of different analyses that are coming out. But if you think about the kind of base analysis that's coming out, it's not so different than what a first year analyst is going to give you.
And you're sitting there listening or taking insights from a first year analyst, you're going to end up finding that you're basically managing an army of, or an infinite army of new grads, each of which you need to check their work, each of which you need to make sure that it's actually providing you the information that you need. And you're going to spend all of your time doing that. And that's not where your value add is, that's not where your differentiator is. And so.
[00:18:54] Speaker B: Sounds onerous, honestly.
[00:18:56] Speaker C: Exactly, exactly. So no one wants to be checking every single bit of work that's coming from their human analysts or their AI analysts. It doesn't really matter. But the key to really solving that problem is you've got to find a way to embed the context that differentiates your strategy from the market in a way that fundamentally gets built into every single decision that is made. Every single decision that's made. People have been doing this for a long time as they're training their human analysts, but also in the way that you're training your AI analysts. The benefit is that the AI analyst will just continue to use it over and over and over again. And so if you can really get that context, as we talked about, like the thesis behind your positions, how you read market moves, like how you read your own book, each of these different components, when they're built into the fundamental analysis that people provide you, or that the AI analysis is providing you and it's embedded, that becomes a huge differentiator and you really end up with a team that's operating at just a much higher level and is able to give you the types of analysis and the insights that are going to drive faster iteration, faster trades into the market, faster consumption of every single piece of information that previously might have taken weeks to get into your strategy. Now you should have it basically built into your strategy almost in real time, obviously with the overlay of the most senior people, making sure that that is really driving the types of change in the market that the insights are showing you. But that the point is that I think working at that level where the context is embedded is core to making any data work at this stage, especially in the age of AI.
[00:20:44] Speaker B: And it sounds also like this, this context idea that you're talking about is kind of like that's the moat. Like that's the DNA of your firm. That's what makes your, your talent, right? Your analysts, your process, all that stuff that makes your fund, your fund, the way you work, the way you research, that's like the, that's the moat, right? That's, that's the differentiator. So no matter how good of a model Is it doesn't matter if it doesn't have that about you. Right, exactly. Just generic LLM answer. No. You want the context of how you think about the world, how you come to decisions. That's the stuff.
[00:21:21] Speaker C: Yeah. You, you absolutely. The context of that. You want the, the years of subject matter expertise that it's taken to build the investment processes at the highest level, truly embedded. Because when you have that.
[00:21:33] Speaker B: Yeah.
[00:21:33] Speaker C: At that point that's where you're going to be able, that's where you're going to be able to beat the market.
[00:21:38] Speaker B: Yeah, that's awesome. And beat the market is one thing. These managers are also thinking how do we even keep up with like a two Sigma or a millennium? Right. Like these big shops, like a 0.72 tick, right.
They're so massive and they have so much resources, they're outspending smaller managers.
Factor of like a thousand. Right. Like let's say something crazy. And if the story is democratization, how does AI help them even try and compete? Right. So let's say they do everything right. Let's say they take your suggestion and they codify their kind of like institutional knowledge or whatever the process right into the context and they feed that into their AIs. They can't hope to replicate what like a large pod shop has. Right. Can they, can you talk a little bit about that democratization?
Does AI help them catch up or are they doomed to just kind of like be forever in this backwater left behind by all the bigger shops?
[00:22:38] Speaker A: I think it's important to understand that they're not competing for the exact, in the exact same kind of sandbox. Right. If I'm a multi manager platform and I've got 45 portfolio managers and I have to manage across the netting risks and all the things to do that very different than a single manager platform. Right. That's focused purely on one aspect of what they're doing and can be really, really good at that. Right. And I think that that's where the markets have afforded, Right. The ability for very different types of investors to play and to be able to understand the what type of, what is the, what is the edge that they're going to need to be able to be successful. I remember having a conversation with a fundamental PM that said I can't compete with these sort of quote unquote pod shops, so to speak. They have all the data they understand kind of like they can better handicap where the next quarter is going to be. I'm looking ahead six to 12 months and I'm looking at what's mispriced, I believe based on the next call, even 12 to 18 months. I'm looking for kind of companies too. Not just any company, but kind of company that there is some sea change that I can get behind in the next six to 12 months, that current quarters are going to be priced in. Right. Because everyone's looking a current quarter, not looking out a year. I can play out in that longer horizon and make sure that I invest in those types of companies. So the point that I'm making is that there's a lot of different ways to make investments, as we know in the marketplace. And some do require a bundle of analytics and data to be able to manage at the 95% IDO or whatever it is, the 85% IDIO exposure that you need. And others are more about how are you truly positioned right. In the types of names you have high conviction in and can concentrate into those names.
So say, how do you level the playing field between all of them? Iran's point is very clear. It's about making sure that the context, that the differentiation of your business and knowledge is aligned with the data, the analytics, the risk management processes, everything that kind of supports that is done in a way that doesn't take you weeks to make a decision.
Right. On a particular investment or on a particular choice of how you're going to position your portfolio. And some of these things do take a long time to get there because of the nature of having to go back to the data, interrogate the data, ask the analysts, go back again. AI speeds that process up. It doesn't solve it completely. And that's the important thing about judgment. It still requires judgment, but it speeds it up where you're not. You don't have to be waiting for a long time to make these decisions.
[00:25:05] Speaker C: Yeah. I also wanted to jump in here and add that I find it, I find it interesting, right, that there's also kind of like a secondary thing that's happening with the larger firms that we're noticing, which is there's a lot of compliance bottlenecks that they're running into as they're trying to, to bring in these types of new technologies when they're running into these compliance bottlenecks. Which makes sense, right. Like if you think about what technology teams and how technology teams adopted it. Right. Like there isn't a significant. At least from an investment management standpoint, it wasn't a significant intellectual property risk. Right. They wanted to get developers coding faster, product managers, shipping prototypes, roadmaps, just Iterating at a really fast pace. I mean, that was great, but really, if you think about where the fund's intellectual property sits, that sits with the investment teams.
And so making sure that the bar, right, was obviously, it's far higher there, making sure that the, that all of the information that's coming in it, like, tracks the different compliance cycles. Like, that was very important.
And what we're seeing is, we're seeing that that's. It's taking quite a bit of time for those types of firms to get past those hurdles. So there's actually an interesting opportunity like right now for these firms to be able to take, for firms that are kind of like smaller in nature, more nimble, like, able to apply technology and tools faster. Like, these are the firms that right now can take that advantage, get, get their, their context built in to the latest and greatest AI tools, making sure that they're leveraging the latest and greatest models, really building in the right structure, which I'm happy to talk about, but really building in the right structure to make sure that people feel like the outputs that they're getting meet their bar or exceed their bar significantly and just iterating with that quite quickly because that's an area where right now they can have an edge relative to a lot of those large firms.
[00:27:01] Speaker B: What you're describing is classic. And it's not just in the buy side. I think it's this disruption, this opportunity for disruption is happening in every single industry in the world.
Anything that AI touches, this is what's happening right now. And it's a blessing and a curse being so big and having so many resources, like you say. Yeah, like it's going to take for these big firms a while to change the workflows of all the pods and all the underlying processes that they have. There's compliance issues, there's just, there's the gravity of what we already have kind of already works. We're already doing great. Like, should we even change anything versus a nimble manager? Oh, new AI tooling. Let's try it. So, so can we, can we dig into that a little bit? Because I think this is where most of the opportunity is and this is where, where it's very, very exciting for, you know, the 500 million to maybe like a 3 to $3 billion shop that's like in that middle. What. How do they do it? Right, so, and you feel free to bring in Omega point, please do like, hey, here's how you think about it. Here's how the structure should be. Avoid this don't make a mistake here.
[00:28:02] Speaker C: Right.
[00:28:02] Speaker B: Don't do something stupid. Like, don't. Don't just treat AI like a chatbot. Right. You really gotta treat it as an agent.
Tell me how they can do this and how fast they should be moving and maybe also working, like, what mistakes should they avoid? Give me kind of like the most important pointers.
[00:28:21] Speaker A: Sure. I mean, Aran, I'm happy to kind of jump in more from the client perspective because we're obviously working with many of them to get that adoption. And, and there's some that the benefit that we have is that we're seeing folks that are way, way, way down the curve that have been implementing this now before, even, like, Claude was a thing. Right. And then there's ones that are sort of obviously in the process of still trying to understand where this all fits in, into their investment process. But what I would say is that there's probably like two or three key things that we're noticing that are very, very important to begin with. Right. The first one is that trust has to be established between the AI processes that are generating the outputs and the PMs or the users. Right. That are using them. And you cannot skip that process and just assume the AI can do things for you. Oh, yeah, anthropic or just Claude will be able to solve this problem for me. Oftentimes I talk to people and they said their first experience with Claude was amazing. Their second experience with Claude was disappointing. Why? Because the second time around, they didn't get the same answer that they got the first time around, and they thought that they should get the same answer. But it turns out that it inferred something slightly differently, like, oh, wow, I have to think about this thing called determinism.
When do I ever. So all of a sudden, when you have to put the AI together, you realize that you need to be able to build that sense of trust where the data that comes back has to actually be evaluated and is coming from a provenance. Right. That can be repeatedly asked again and again and receive the same answer. I think that that's step one. Right. That you have to establish. And again, the core is naturally, if you partner with folks like us, we've done a lot of that hard work to make sure that the provenance is there and that when you ask an agent a question, you're going to get back the same answer because it's actually using the same underlying calculation and not inferring it. But that's. That's step one. That's a really important piece, establish the trust and establish the underlying kind of data that you can trust to go forward. The second piece that Aran talked about that's really important is you need to put together the context of your investment process, right? And again, none of these things take a long time if you understand how to build it. Right? But the idea is that you have your investment process. You have everything from how you do research to how you think about position sizing, how you think about monitoring and managing risk, to how you think about what type of hedges to put on, how to communicate with your LPs. Every piece of your investment process and how the. How that's all structured can actually be codified within an AI in a very discreet manner. You don't have to just say, AI, here's my investment process. Read it, and you're good. You actually want to be able to create discrete small pieces, building blocks, right? AI can understand because then it can assemble these together when something happens and be able to learn from them. So those two things become foundational to what you're working on, right? Because then you're able to say, I have the data that I like, provenance is good, and I have all the building blocks built. Then once you have that, the foundation has been set to start to. You actually starting to do some really cool things, right? To be able to establish things like triggers, right? Oh, when X happens, AI, I want to be either notified or told, what do you think makes sense? And I trust that it's going to be able to use all the building blocks that I created under the hood, right? To be able to do that. So that's sort of how you start, at least from my perspective, our perspective down the journey. And again, working with folks like us, you could accelerate that process a lot than having to build it from scratch on your own.
[00:31:44] Speaker B: That's great.
Iran use cases. That's going to flip that. Aha. Switch with fundamental managers. Like, for instance, triggers. That's an obvious one, right? Boy, it's hard for a human being, especially with a small team, to monitor everything all at once. But you get into AI agents, even if it's analyst level, like, still you could get it. Is there anything else that you could point to Iran like that, you know, when you say it like a PM or an analyst would say, oh, dude, I could use that.
[00:32:14] Speaker C: Yeah. I mean, listen, I think that it comes down to compounded learning. That is one of the most important things that an AI can do, right? You've shared a piece of information with that AI. Once it's surfaced an insight for you and you've had that type that conversation with it. You want it, you want to incorporate that into its context the next time that trigger fires. Right.
Otherwise you're in a situation where it's kind of like you're telling the analyst or you're telling your team the same thing over and over again and you're not getting the results. So you want them to constantly learn. I want to distinguish that from something that people talk about, which is training on the context or training models based upon it. I think that those are two very different things. And actually I would caution people against the training aspect. That is something that is more of a compliance hurdle that's going to cause friction within your organization to actually get AI working. So one of the things that a megapoint does, it doesn't train on any data that customers provide. But we do provide customers the ability to really have their agents get smarter through compounded learning by essentially ensuring that the context, when you provide it additional feedback, we say, hey, you've told us this before. Let's make sure that that's a part of the context every single time we think about how we should answer one of your questions.
[00:33:36] Speaker B: Again, I'm so glad you drew that distinction. Right. Because training is kind of like, it's the boogeyman. Like in the investment management industry where everybody thinks they have a very proprietary process and it's just truly mine. You don't want some model to be trained on that, God forbid. Right. Because like, what if someone else then just uses our proprietary thing? So yeah. So you kind of want to be careful. Right. My other question was around like, well, why don't just hedge fund managers just use Claude? Because it's got all this stuff, right? It's got great skills. You can load a finance skill, you can make your own skills, you can kind of feed it your own data. You could begin to do that. Right. Then I guess. Can you guarantee that Anthropic is not training on your data? Not. I don't know. They say they don't. They say they don't. Right. And for some people that's enough. But I, I could see like some hedge fund managers are going to say, well, that's not exactly enough for me. So where, where is that difference? Make it black and white, make it really, really simple. Like, no, guys, like, use Claude for some things. Sure. And skills and, and all that stuff and it's going to get better. Yes, but you still need Omega Point if you want to take it to the next level. Can you make that distinction super, super clear?
[00:34:47] Speaker A: Sure. Again, I'm happy to. I spend a lot of time with our customers when they ask these types of questions and they're trying to ascertain first of all, how do I use my own internal Agentix system. And now there's different, as you know, there's different ways of using claude. You don't have to use the entire kind of CLAUDE package. Right. Cloud code, cloud cowork, if you decide not to, you can use CLAUDE in other types of kind of localized environments and so on and so forth. So there are definitely different ways of leveraging the AI. And I think compliance teams get involved in a lot of these and IT teams get involved in how they sort of build out and make those sort of decisions. I think the important things to understand about how to make even the decision to go with an AI in a way that's safe. Right. That's safe for you to be able to understand either the training associated with it, if it's being trained or it's not being trained and how you can actually trust that are giving you the results that you're looking for. And this is where you asking about the black and white aspect of it is going back to the notion that can you trust the output that it comes from it? Right. Based on understanding of the provenance.
[00:35:52] Speaker C: Right.
[00:35:53] Speaker A: Where is the output actually coming from? Right. And is it deterministic in the sense that in our world of finance most things can be computed not generated. Right. Meaning computed, not generated, I think is the word that you might be thinking of.
If it's generating a bunch of things for you and reinventing risk again and again, that's not a great idea. Right. Maybe you do want to do it, but that's a research project. That's not what you're going to be doing in your day to day, your day to day. Your process is going to leverage well known and built tools and you want the AI to orchestrate and to route to the right places to make these decisions. And all these are computations that are well known and grounded in real math.
So if you can guarantee that that actually happens. Right. Then you have a lot more trust that the AI is able to do that. And that's where I think companies or systems like ours help you get that level of trust. It's all grounded in those kind of pieces. The second one that's important for black and white involves also a piece that's information, which is there's still data that comes out from these AIs and there needs to be a level of subject matter expertise for two, make sure that that data has this particular interpretation context. Right? Yeah. I'm sure you know this. You can ask a bunch.
[00:37:07] Speaker B: Yes.
And that's that context that Iran was talking about. It's like, what does it mean for me? What does it mean, like, for me in the context of my portfolio, in the context of my themes, with all that other stuff that I already told you I care about, I don't care about, what does it mean for me? That is like, Claude, this just does not know, right? That's just some generic. Right? Yeah, yeah. So this is great, right?
Yeah. What are you most excited about and where can this democratization of space can accelerate tap dance to work today?
[00:37:38] Speaker C: Yeah, listen, there's just a sea change in the way that people are working right now and that's extremely exciting. Right. This is like something I've never seen before. Right now people are looking at how they were doing things even three, six, nine months ago, and they're saying, hey, there's a 10x opportunity here. Right. And it is about incorporating. How do I, how do I rethink this in this new world where frankly, I can leverage all of this investment that's happened in AI architectures and LLMs in order to just make my process so much more productive and to rethink how I engage with the markets. And I feel like every single conversation I'm having with clients is shifting in that direction. And ones that are forward thinking, man, are they seeing an advantage. It is incredible to just sit with them and work with them and help them really achieve kind of even the next level. And obviously that's a lot of what we do at Omega Point and we make sure that we're like all the failure mode. Right. That happen at that edge.
[00:38:44] Speaker B: Right.
[00:38:44] Speaker C: We're already building into the systems that we have to make sure that people like every single client that we have can really get that benefit.
[00:38:54] Speaker B: It's amazing.
[00:38:57] Speaker A: Yeah.
[00:38:58] Speaker B: What an opportunity, what an interesting side to be working with technology at the cutting edge and the buy side. Fantastic. You guys have been absolutely great. Our listeners, if they wanted to reach out or learn more about Omega Point, what's their best, best source? Should I go to the website? Should I reach out on LinkedIn? Where can people find you guys? Absolutely. So it's been absolute pleasure having you both on the show. So for everyone who's listening, the links to both the website and the LinkedIn, along with the BIOS where you can find Omer and Iran, are going to be included in the description under the video. We hope you enjoyed this episode.
I thank you both so much for being guests at the Momentum Podcast, and for everybody who's tuning in, thank you for making it all the way out to the end of the episode.
[00:39:38] Speaker A: Thank you.
[00:39:38] Speaker B: We appreciate you. Give us a follow and we'll see you out there. Thanks a lot, you guys.
[00:39:42] Speaker C: Thanks a lot.
[00:39:45] Speaker B: Thanks for tuning into Momentum. If you found today's episode valuable, don't forget to follow the show and share it with your network. This helps us a lot. To explore how AI can help your firm, visit Acadia im. I'm Stan Altschiller and I'll see you next week with more practical insights to help turn data into growth and decisions into results with the help of AI.
