E326: What Happens When AI Starts Replacing Analysts?

0:00You built Model ML to solve your own problems within your family office. What were those problems? So when Anz and I, who's my brother, we sold the second company, we were still pretty young. Anz was, I don't know, 21, 22. I was maybe like 26, 20, 27. We made a bit of money and we decided that we wanted to invest our own money for a while. So we decided on a few kind of asset classes, strategies. We hired a few people to predominantly do the investing. And then we spent a lot of our day writing software to make that process better. Before we knew it, that was maybe 21 coming into 2022. 2023 coming into 2024, as LLM started to really be useful in these sort of environments, frankly, the product just got better and better and better.

0:43And before we knew it, we just had too many people asking us whether they could use it and they're willing to pay for it. So the story goes that we called my mom, asked my mom if we could build a third startup and she said no. And so we started it the next day. that. Give me a specific low hanging fruit that investors are using Model ML in order to solve their everyday problems. A classic is just like reporting and monitoring in general, right? It's the classic problem. Even when I think about this back at the family office, we were making on the venture or startup size, maybe like 25 investments a year, right?

1:15And we would receive updates in like every single format. Sometimes even now it's like a website link. It's like, you've got a website link, you've got a Notion page, you've got just a bunch of documents, you've got it in the body of the email, you've got it in an Excel file with multiple tabs, et cetera. And really what you want to do is you want to consolidate that down and bring that into your systems in your format. It's somewhat baffling to me that a lot of these tasks are still being done manually, frankly. That is a classic example of something that should be automated. We're not in the world of like 100 % automation.

1:47I think we'll get there, you know, probably 12, sort of 24 months away of a single task being close to 100%. You know, anything above 60 % automation, we really focus on. So, you know, it's not about getting, you know, to pixel perfect to the end. It's really where can AI be most applicable in that specific workflow today? And a classic one is something like reporting. And 2025 was supposed to be the year of agentic AI. Now people are saying 2026. You're one of the only agentic AI companies on the planet that has scaled to use over a trillion tokens with open AI. why have you been able to solve agentic AI in a way that others have not?

2:23We relaunched the end of 2024, and we raised about 100 million across a couple of rounds in our first 12-ish months. The business has been growing, right? And looking back on it, we often think, okay, where did we differentiate? And to us, it's quite clear. These chat-type interfaces, think ChatGPT and Anthropic and others, they're great. Don't get me wrong. They're absolutely fantastic. They're very much going to change the world. But if you think about the complexity of work that actually goes on in these types of organizations, they're going to have a ceiling. So if you think of any relatively complex workflow that you've done in the last few months, can you solve that in a chat or Q &A type environment?

3:04If that's resulting in a 200 page PowerPoint or a 50 tab wide Excel workbook, are you going to be able to solve that as of today in a chat interface? The answer is almost certainly no. And we kind of knew that up front. And so we came into the market with a slightly different perspective where it's like, look, the chat type interface is great and we have that and it's useful. But really, I think what firms want is they want true workflow automation. But also it's much easier for us to sell to them and easier for them to procure as well, right? Because they can actually say, okay, well, here are the three or four things that we want to automate.

3:39Can you automate those things? Well, if we can, yes. And it's very clear. The difficulty with a chat type interface is it's very hard to quantify. Whether that's delivering additional insight or productivity, it's very difficult to quantify. With workflow automations, it's just a lot easier. How should investors, GPs, or LPs think about agentic AI and where could they apply it in their day-to-day? The important thing is, where are we today? But where is this heading? So today, let's make no mistake about it. you know, these application layer products, whether that's in finance, legal, healthcare, whatever, you know, they are productivity tools for the most part.

4:16They're, they're giving you, you know, an additional layer of efficiency in your businesses, right? The question is though, particularly if we think about this from an investing standpoint is when is this going to be able to deliver a level of insight that wasn't possible pre-AI? That's really the question, you know, because at the end of the day, productivity is great, but, but it's all about that alpha from an insight perspective. And in our view, there is absolutely no doubt that 2026 is going to be that year. We think 2025 is the productivity year. 2026 is the year where we're actually going to start to see these systems deliver insight.

4:49That wasn't possible pre-AI. Now, if we got to play that back a little bit, that's not going to be today, no insight, tomorrow insight. It's going to be slightly more incremental than that. And I'll give you a quite specific example. One of our middle market private equity clients, it's a European client. They're absolutely fantastic. They've pretty much automated 80 % of their IC memos, their IC paper, right? A lot of that is going to different data sources and really just data retrieval, a bit of reasoning and producing that in a format that they're used to digesting. Graph tables, charts, logo in the same formats that they would digest the information from, you know, going to the data room and going to CapIQ and going to pitch book and so on, all the areas of information that you would normally go to.

5:32But there's two or three pages in that now that are not just generated by AI, but it's kind of like the AI's opinion, right? And these are things like, you know, the AI's opinion on the overall management team, based on your historical investments. We've noticed that there's a lack of experience over here. There's a lot of experience here, for example. Now, as of today, they glance over that in their IC meeting. It's kind of like, this is interesting. We spend five minutes on it and we move on. But one of the things that's clear to them and really clear to us is the importance of those two or three pages is only going one way.

6:04In other words, the AI opinion is only becoming stronger and stronger and stronger. And so I think that's why it's super important that firms, you know, it might not be perfect today, right? But you've got to embed this into your culture and these systems into the way that you think as soon as possible, because that future insight is only going to be unlocked by doing this today. Tell me about that. Why do you have to prepare today for insights in the future? I was thinking coming into this conversation, what would I advise, you know, irrespective of what we do, what would I advise firms to think about today?

6:34I really think about data. I think things like trying to transcribe calls is a great example of like, you know, if you think of what these systems are going to need in future, the more data that they have, particularly now because they are a sort of data structure agnostic, whether they're calls, emails, files and folders, structured data, it doesn't matter, right? You want to try and capture as much data as you possibly can as part of the investing process, I think that's important. The second part of that question is, this is more of a cultural shift than anything else. I think as we've thought about, so I should say, our customers are about a third, say asset management in general, a third in the largest consulting firms in the world and a third in banking.

7:13There are thereabouts. There's a few other others in there now, but there are thereabouts. Now, the consistency across all of them, right? So not just on the buy side, the consistency across all of them is this is clearly not a technology problem anymore, right? This is becoming more and more of a cultural change and a structural change, you know, as to how you think about the organization, how you think about AI from a cultural perspective, but that takes time. And, you know, I really would encourage firms to just start not overthink that initial process and start. One of the hardest things of investing is seeing what's shifting before everyone else does.

7:46For decades, only the largest hedge funds could afford extensive channel research programs to spot inflection points before earnings and to stay ahead of consensus. Meanwhile, smaller funds have been forced to cobble together ad hoc channel intelligence or rely on stale reports from sell-side shops. But channel checks are no longer a luxury. They're becoming table stakes for the industry. The challenge has always been scale, speed, and consistency. That's where AlphaSense comes in. AlphaSense is redefining channel research. Instead of static point-in-time reports, AlphaSense channel checks delivers a continuously refreshed view of demand, pricing, and competitive dynamics powered by interviews with real operators, suppliers, distributors, and channel partners across the value chain.

8:29Thousands of consistent channel conversations every month deliver clean, comparable signals, helping investors spot inflection points weeks before they show up in earnings or consensus estimates. The best part, these proprietary channel checks integrate directly into AlphaSense's research platform trusted by 75 % of the world's top hedge funds with access to over 500 million premium sources. From company filings and brokerage research to news, trade journals, and more than 240 ,000 expert call transcripts. That context turns raw signal into conviction. The first to see wins, the rest follow. Check it out for yourself at alpha-sense.com slash how I invest.

9:07It's interesting because a lot of people use this analogy of data is the new oil, but no one goes upstream of that and thinks about how do we capture more data? You drill in the ground to get more oil and you have to capture the data. If it's truly oil, why aren't you capturing more of it? And why aren't you creating the schema, the processes, and the cultural aspects of becoming a data-driven organization? The reason I'm always using that example, sort of cool transcriptions is because it's kind of in a way obvious, but less obvious, right? These models, they're incredible at interpreting videos, um incredible testing audio text it doesn't doesn't matter right and so it's almost like regardless of how you're thinking about ai in a short mid long term start capturing that data we did some analysis around you know what actually goes into the decision making process and where that information comes from it's something like 70 of all information is coming from calls and meetings right so you want to try and capture as much as that as you possibly can again this is this is less relevant to us.

10:04It's more just our view on like, okay, outside of adopting a vertical application, which clearly seems like something that firms should do, what else can they start thinking about doing sort of agnostic of that choice? And it's just capturing more data where they can. The last time we chatted, you gave me this shocking statistic that you believe that companies will experience 55 % productivity gains in the next 12 to 18 months on the backs of AI. Where are these productivity gains specifically going to come from? So I was saying, well, what even qualifies us for making that statement? A lot of what we're doing as we're selling into firms is we're building business cases, right?

10:41So we have to deeply understand where AI is applicable today. The big thing that really no one saw is the progress that happened with reasoning models. You know, reasoning is effectively a technique. It's very difficult to forecast progress in underlying techniques. Now, forecasting progress in the underlying models, you know, is theoretically, you know, that's a lot more predictable, but techniques are more difficult to predict, clearly. And so as we look at that and we look at all the data and really looking at where the time is going, more importantly, particularly the more junior members of the team, you know, we're looking at a 50 % efficiency, certainly before the end of 2027.

11:20But that could be a considerable amount of time before that. And I don't think that's because of the technology. I think that's because of the adoption. Part of your business processes is you go into organizations and you look for productivity gains that they could have using agentic AI. What's the exercise that you go through? How can an organization figure out where they have the most productivity gains? We do an entirely free, you don't have to sign up, discovery phase, right? So, you know, we will work with a customer and we will identify where we think AI is most applicable today beyond just a chat-based interface, as I said, workflow automation.

11:58So we do that over a two, three, or four-week period. And really, come the end of that period, we're targeting a number of... And we don't really look at workflows unless it's a 60-plus percent efficiency gain on a single workflow level. The reason we do that is because we really want to focus on short-term ROI to deliver value quickly. But we do that now with all of our customers. perhaps it's a dumb question but is this a soft is this a software process that's running on machines are these virtual machines how do you actually implement these systems the user just logs in presses a button and the report is generated how does that mechanically work so if you kind of picture it the way it would work is we have an excel type interface so think of you know just excel but powered by ai you upload a document and what we're doing is we're deconstructing that document into its individual components and we're making you know the model is making the best guess as to where you would get that information from, from documents, from fact set, from verbal filings, from wherever you may normally get that information from.

12:51The user would come in and confirm, they would click save, and then that workflow would exist. And they can deploy that workflow to just themselves, to the rest of the team, to the whole group. And then they can obviously, as you go, you can make alterations and then someone can make a copy and adapt that workflow to them as a specific user if they want to. Is it fair to compare you as an enterprise nice open cloth of sorts? And how do you look at open source projects like open claw with competing with what you're doing? Quite important. That has become very clear in the legal AI space, I think, you know, I think it's like 80, 90 % of the top 100 law firms now use either Lego or Harvey.

13:24And I think finance is about 12 months behind. So I think we can learn a lot from what happened in that market. And if you look what happened there, it's really interesting with us is the entire product, you know, that's the bottom of the agendistic system through to the UI is designed with the customer in mind, right? And that last mile delivery to the customer, really from what we're seeing is where the impact is. So it's not what the agents are necessarily going and gathering, it's how that's presented to the customer. And it's very specific to that customer base. Exactly. And it's just, it's the design of the overall system, right?

13:57So it's like, you know, if you think about the agents, it's what data you integrate with, the way that the agents communicate, what language do they communicate in? Do they understand financial concepts well? Where is the context coming from? So this, you know, from an agentic perspective, there's lots of finance-specific things that go on that are really, really important. Again, you've seen that in the legal AI world. But then it's the UI. How is a user interacting with these systems? You know, coming back to this point of, can you build a 200-page, you know, PowerPoint presentation exactly as you would have done before?

14:26Graphs, tables, charts, logos. You know, think about the complexity that goes into these types of outputs. Think FDDs and CDDs and so on, right? Can you do that on your phone in ChatGPT or in Claude? No. And so you have to, it's the agent system all the way through to the UI and with user interacting with these systems that becomes really, really important. You mentioned it. Legal tech really pioneered this professional use of agentic AI. What can you learn from the Harvey's, Legoras of the world? And how do you apply that to the fintech? Support for today's episode comes from Square, the all-in-one way for business owners to take payments, book appointments, manage staff, and keep everything running in one place.

15:06Whether you're selling lattes, cutting hair, running a boutique, or managing a service business, Square helps you run your business without running yourself into the ground. I was actually thinking about this the other day when I stopped by a local cafe here. They use Square and everything just works. Checkout is fast, receipts are instant, and sometimes I even get loyalty rewards automatically. There's something about businesses that use Square. They just feel more put together. The experience is smoother for them and it's smoother for me as a customer. Square makes it easy to sell wherever your customers are, in store, online, on your phone, or even at pop-ups and everything stays synced in real time.

15:41You can track sales, manage inventory, book appointments, and see reports instantly, whether you're in your shop or on the go. And when you make a sale, you don't have to wait to get paid. Square gives you fast access to your earnings through Square checking. They also have built-in tools like loyalty and marketing, so their best customers keep coming back. And right now, you can get up to$200 off Square hardware when you sign up at square.com slash go slash how I invest. That's S-Q-U-A-R-E dot com slash go slash how I invest. With Square, you get all the tools to run your business with none of the contracts or complexity.

16:16Run your business smarter with Square. Get started today. If you've been considering futures trading, now might be the time to take a closer look. The futures markets has seen increased activity recently and Plus500 Futures offers a straightforward entry point. The platform provides access to major instruments including the S &P 500, NASDAQ, Bitcoin, natural gas, and other key markets across equity indices, energy, metals, forex, and crypto. Their interface is designed for accessibility. You can monitor and execute trades from your phone with a$100 minimum deposit. Once your account is open, potential trades can be executed in two clicks.

16:51For those who prefer to practice first, Plus500 offers an unlimited demo account with full charting and analytical tools. No risk involved while you familiarize yourself with the platform. The company has been operating in a trading space for over 20 years. Download the Plus500 app. Trading and futures involves the risk of loss. This is not suitable for everyone. Not all applicants will qualify. When you want more, you start your business with Northwest Registered Agent. They give you access to thousands of free guides, tools, and legal forms to help you launch and protect your business all in one place.

17:22With Northwest, you're not just forming an LLC, you're building your complete business identity. From what customers see to what they don't see, like operating agreements, meeting minutes, and compliance paperwork, you get more privacy, more guidance, and more free resources to grow the right way. Northwest has been helping founders and entrepreneurs for nearly 30 years. They're the largest registered agent LLC service in the US with over 1500 corporate guides, real people who know your local laws and help you every step of the way. What I love is how fast you can build your business identity with their free resources.

17:54You can access thousands of forms, step-by-step guides, and even lawyer drafted operating agreements and bylaws without even creating an account. Northwest makes life easy for business owners. They don't just help you from your company. They give you the tools you need after you form it. And with Northwest, privacy is automatic. They never sell your data because privacy by default is their pledge. Don't pay hundreds or thousands of dollars for what you could get from Northwest for free. Visit northwestregisteredagent.com slash invest free and start using free resources to build something amazing.

18:25Get more with Northwest Registered Agent at northwestregisteredagent.com slash invest free. Use case. That last mile delivery we talk about is really, really important, right? So when I talk about last mile delivery, I'm sure you've heard of this concept of FDEs, forward deployed engineers, that's like booming. That's the concept of having people literally work in the offices or from the offices of your customers for a period of time to not just drive adoption, but to deal with onboarding and also to innovate really, really quickly. I think that you saw a lot of that in the AI. I think that is just as, if not more important in finance, again, because it's not about just the adoption and driving the cultural change.

19:06And it's those really, really quick feedback loops. One of the things that we've become very well known for if we look at the competitive landscape is, if we look at our nearest competitors, we're maybe built a better product in a quarter of the time with a tenth of the capital. Now, I know that we've raised a lot recently, but before that, we're the tenth of the capital. I think that's really, really important because if you look at the legal AI world and our world, our customers are not betting on our product today. They're betting on where our product is in 12-month time. So, you know, for all intents and purposes, they're betting on the team and the team's ability to ship, you know, the best of the best and be at the forefront consistently.

19:42Right. And so I think, you know, with this forward deployed model that you saw in the league, I space a lot, you know, we're doing it. What's actually happening there is we're taking feedback in real time and we are altering and approving the products in real time. And let me be specific. You know, we have scenarios where someone will be at a customer's office. They will receive product feedback in real time. they will write all to the code base in real time, make that alteration, do the pull request, and it'll be in production within half an hour. And some may say, well, that doesn't make it.

20:12It's great, but is that really needed? What about the day after? But the thing is, in this world is, these systems are improving so quickly and the competitive landscape is changing so quickly that really our moat as a business is the speed at which we can learn and therefore share products. Does that not limit your scale? how do you go about scaling a services-based? Maybe this is Arnie and I being a little bit English, or maybe this is the Y Combinator way kind of installed in our mind. But our motto here is like fewer happier customers, right? We're not a business that's going to over-promise and under-deliver, right?

20:47We're a business that when we say we're going to do something for a customer, we do. And if we don't, I think it hurts us emotionally, right? So I think our view on the market is, you know, these demos that you see all over LinkedIn and so on, they're great. But the things that really matter, you know, if we fast forward 12 or 24 months time, you know, it's going to be that sticky revenue. Sticky revenue is ultimately going to come from, you know, how happy your customers are with your product. So sure, look, we can't take it to the extreme, you know, do things that don't scale. But at the same time, we just believe that if we deliver to our word in a market where reputation, frankly, is everything, is the single most important thing.

21:27Now, you can still scale quickly. We've raised, as I said, about$100 million in our first 12 months. So we're very well capitalized to do so. The team's gone from 20 people to 100. I'm sure we're allowed another 100 in the next few months. So we're scaling to accommodate for that. And sure, it might mean in the shorter term, our margins are slightly less as a business. But ultimately, we think if we do well in that first month, three months, six months, and 12 months, we can have these customers for a very, very long time. Talk to me about the early adopters in the GP and LP space. What are the characteristics of these organizations and where are they adopting it internally?

21:59It's interesting you say about characteristics because I think that comes back to this thing of a cultural shift, right? So I think it's really important that there is top-down buy-in, CEO level or CIO level buy-in of these sorts of products or AI in general. general, which now I think is almost a bit outdated. I think that statement was relevant maybe six months or so ago. I don't know of a single firm where this isn't like number one on the agenda or maybe number two, but probably number one. So generally speaking, the best early adopters that we've seen are where there is like true kind of top down buy.

22:39I think that's the first thing. The second thing is we've also noticed it's the firms that think about how this is going to alter the organization structurally, because clearly it will. I don't think it takes a rocket scientists to figure out, you know, you go on chat, GPT and do anything around, you know, any of these AI products, you know, there's, there's a huge efficiency game that's happened and happening, right? So there has to be a structural change. So the ones that are really thinking about things structurally as an overall organization, I think are the ones that are the best editors being, if you want me to be specific that I really mean, well, if, if you take your, your, your more junior members of the team and you're seeing, even if we're pessimistic and we're seeing a 10, 20 % overall possible efficiency game, well, where is that 10, 20 % going?

23:19It needs to be reallocated. And the best firms and the best leaders, the ones that are thinking ahead and they're saying, okay, well, we're going to take this pool of say 10 % of the overall team, and they're going to become our AI experts. A lot of time they're self-identified. We all have friends or people that naturally in their roles are like super interested compared to their peers at AI. They're probably the ones that you want to use because you don't need to write code in this sort of environment, right? So the ones that are actually saying, okay, we need to move these people and your job is to either assess or you know ai tools or build ai workflows in a product like model ml they're the ones that i think are doing incredibly well the last part of your question is you know where do you start you know what areas of the business do you focus on i think this is you know can be on a case by case and so that's where we've been very value add in this discovery phase so i really think like we we do an incredible job at working closely for you know this free discovery phase, helping, you know, whether that's GPs, whoever it might be, you know, identify areas that they should immediately be thinking about applying artificial intelligence.

24:20That's what we've become very good at. Now, what does that actually look like in terms of use cases? Well, for now, it's more of those, you know, inefficient, you know, areas of the business that you can have these productivity gains. As I said, I think in future, that'll become more how you're able to drive insight that wasn't possible pre-AI, but for today, it's productivity. What are we talking about budget-wise? What size of firm do you need to be to hire a firm like Model ML? Or maybe how much do these projects cost? We as a business don't do many kind of like 5, 10, 20 seater type deals.