投资者如何提升专家访谈与 AI 能力：AlphaSense 的 Ryan Fennerty

All right. Hello and welcome to the Another Value podcast. I'm your host, Andrew Walker. Today I have a really interesting podcast for you. I say that all the time, but look, I think this is going to be a specialized one. I think if you are a small fund, well, I should tell you what it is. It is Ryan Fennerty from AlphaSense. AlphaSense is obviously a longtime sponsor of the podcast. So, I know what you're thinking: Oh my god, this is an infomercial. I don't think it's an infomercial. AlphaSense is the provider of AI tools and expert calls to financial firms. I'm a heavy expert-call user, and you're going to hear it. I'm going to grill Ryan on how I can be a better user of expert calls and AI tools as an investor. If you are an investor and you use expert calls or AI tools or both, then you are going to get a lot out of this podcast, in my opinion. And if you are an investor who doesn't use AI tools or expert calls, I'm going to ask you: what the heck are you doing? Get with the times. These are the two most important new tools in investors' toolkit that have developed over the past 10 to 15 years. So I know what you're going to say: it's an infomercial. It's not an infomercial. You're going to learn a lot about how to improve as an expert-call user, how to improve for AI, and how to improve as an investor. We're going to get there in one second, but first, a word from our sponsor, AlphaSense.

Today's podcast is sponsored by AlphaSense. Look, AlphaSense has been a longtime sponsor of this podcast. You're about to listen to a podcast with one of the people from AlphaSense who's going to talk about how you can improve with expert calls and AI. If you've been following this podcast for a long time, you know I believe that, over the past 10 years, the two most powerful tools that have come along and changed things for investors are expert-call networks, which have enabled funds and investors of all sizes to get access to expert calls, and AI, which has enabled all sorts of tools for funds and small investors. AI and expert calls are a match made in heaven. They're increasingly blending together. AlphaSense is rolling out AI-led expert-call tools that let you pair experts with a knowledge-based AI interviewer to conduct high-quality conversations on your behalf. If previously you were limited by, "Hey, I can only do two expert calls a day. Maybe I can't do full surveys and all this sort of stuff," you can have the AlphaSense AI call go and do 100 calls. If you've got the budget, you could have it interview every single McDonald's manager who's willing to sign up for an expert call and get some really interesting insights. I just think it's a match made in heaven. AlphaSense continues to push the edge, push the envelope, and evolve it. I think it's great. You should check out AlphaSense and the AI expert calls. You can learn more at alpha-sense.com/yavvp. And now on to the podcast. All right. Hello and welcome to the Yet Another Value Podcast. I'm your host, Andrew Walker, and with me today, I'm happy to have on from AlphaSense, Ryan Fennerty. Ryan, how's it going?

It's awesome. Good to see you.

Thank you so much for coming on. We're going to hop into the podcast in one second, but quick disclaimer for everyone. Nothing on this podcast is investing advice. I don't think we're talking about any individual securities. We're talking generally about how to improve as an investor and use some interesting tools. Keep that in mind. There's a full disclaimer at the end of the podcast and in the show notes.

The reason I wanted to have you on is that you work at AlphaSense, overseeing the AI tools and expert calls. I've talked about this before, but I think these are 2 of the areas where, especially for smaller fund managers, the landscape has evolved a lot over the past 4 years for AI tools and over the past 10 years for expert calls. I wanted to do an update and talk about all of those for my listeners, so that makes sense. We'll hop into it.

That's great. Just one piece of context for your viewers and your audience: I initially led the expert calls business at Tegus and helped scale it. Then we were acquired by AlphaSense, and now I lead financial services sales for AlphaSense. I bring both the lens of how we were building at Tegus and how that's evolving through AlphaSense, especially as AI becomes a huge part of where the industry is headed.

In addition, AlphaSense is much more of an AI-forward platform to support investors, so I can speak to how we're seeing AI impact use cases in the market.

Your journey inside AlphaSense is like my journey outside AlphaSense because I knew AlphaSense from Sentieo, and then they bought Tegus. It was all about the expert calls for me, and then you've got these burgeoning AI tools. I think we'll talk about this in the podcast, but the expert calls are awesome, and that's what I think about first when I think of AlphaSense. The AI tools are reshaping how expert calls and learning from expert calls are done, and I'm still trying to wrap my head around it.

Anyway, let's start with expert calls. I do a lot of expert calls. I was trying to put a number on it, and I'm going to say 25 a year, but it could be upwards of 50 a year. Of those, I'd say 10% are awesome, 50% are good, 40% are okay, and 10% are bad.

I wanted to frame this conversation around improving expert calls: getting that 10% that are awesome to 30% and getting all the bad ones out of there. That's my overall framing and thought process for expert calls.

Let me start with this question. If someone is listening right now and wants one takeaway—if they wanted to say, “Hey, Ryan taught me one way I could improve as an investor using expert calls”—what is one takeaway that someone could have to improve their expert calls?

At the risk of generalizing, knowing that many different people use expert calls for different, discrete purposes in their investment process, the number-one thing I would say to shift toward having more satisfying expert calls is to approach them through the frame of, “I am testing a hypothesis or a thesis, and I want a credible thought partner to think through that and the second-order implications.”

I think that's where you find the best expert calls. They have a goal and something they're trying to validate or invalidate, and they have enough structure to allow for that to happen. But they also have enough flexibility for you to probe and go deeper.

Anyone who's ever used an expert transcript library and seen some of the expert calls has thought, “That was a great expert call.” They kind of follow that arc; they're flying at the right altitude. Some people come in really trying to say, “I just want data that corroborates this thing I'm trying to test,” and then they come out frustrated that the expert was evasive or gave ranges that didn't make sense.

I'd say the number-one thing is to frame expert calls as being really well utilized for humans who can help you think through a hypothesis you have and really help test your thinking on that.

One thing, just on having a hypothesis: I am a generalist in most sectors versus an industry specialist. How should generalists be thinking about using expert calls versus industry specialists?

For me, I might go in and my thesis might be, “We're recording on February 9 or 10. Software stocks are getting destroyed. I want to talk to someone about this company and how AI is impacting software.” Whereas an industry expert might say, “I already know how it's impacting software. I want to talk to industry people about, in real time, how their spending is changing.”

How do generalists versus specialists differ when they're using expert calls?

I think that's a really fair question. Here's what I would say: zoom out, because one of the things to consider is how this is all changing, given how dynamic the space is.

In general, a lot of the work that used to happen around just getting up to speed and getting smart—first-order questioning to get triangulated on things—has moved to the expert libraries, where you can see what others have done. That's not always true, but a lot of the work that used to go to expert calls to do that has moved to, “Let's look and see what's on these libraries and who else is talking about this stuff.”

Where a lot more of the effort has gone is toward much deeper questions around an investment thesis or the drivers of a company. I think that's where we're seeing a lot more of the behavior on expert calls. Instead of saying, “I'll go talk to 10 people just to triangulate on how industry structure works and big-picture trends,” a lot more of where we're being utilized is the latter stuff that I talked about.

Every interviewer comes to the thing with a bias. I'm an AlphaSense user, a Tegus user, and all these things. Even though I do a decent number of AlphaSense calls, I read a lot more calls than I do live-person calls.

When your best users are making what you consider the best use of expert calls to further their knowledge and all this sort of stuff, what is their blend of the expert calls that they are driving and doing versus transcript usage?

I have a couple of friends who were early users of Tegus and talked about how they shifted their behavior and how they're using it now.

I'd say one of the biggest things that's happened, if you think about where Tegus came into the industry model for doing expert calls, is that we disrupted the price of what an expert call used to cost. It used to be that $1,500 to $2,000 was the average price for a call, right?

Yeah. No, I'm nodding along because I was in consulting and private equity before. You'd do expert calls: “Hey, we're going to spend $2,000 on the expert. It's a private call. No one can see it. We're doing diligence on something. It's going to be between 8 and 40 calls, and this is the biggest part of the due diligence process.” Please continue, but I'm done because I so agree with what you're saying.

So, I see this as the arc that's happened over literally the last 3 to 4 years. That was the state of the industry, and then models like Tegus came in, which basically monetized in a different way through access to an expert transcript library, where everyone's expert calls over time were put there to be searched and read. What that allowed was basically doing expert calls at cost, so there's no margin, and it opened up the market.

I used to be a banker covering airlines. Spirit Airlines expanded the market of people who could actually take advantage of low-cost airlines, right? I think that was one of the giant things that Tegus introduced. A lot of our business was built on mid-market funds that previously couldn't do the volume of calls they could do with us. I'd just say that was the first change: I went from having to be incredibly selective about where I did my expert calls, and doing a lot of them through our own network of people we referred to, to being able to take a lot more of those triangulation calls.

Then what happened is these expert libraries started to form in the market, and there are multiple ones. Tegus has one; there are other ones out there. This is where a lot of the get-up-to-speed work—just understanding, ultimately, market structure, go-to-market model, pricing, operating leverage—a lot of that cursory work got done through the expert transcripts.

But then what you find is people are using those as a stopping-off point for the second, third, or fourth call that they would have done. That becomes the first call because now they can triangulate on a name, see the drivers, and see the other questions people asked. I think the biggest thing we're seeing in the industry, whether you're public or private, is that investing has always been about access to information and then an investment process that gets you to superior investment outcomes.

Access to information for all that insight that was trapped in expert calls has become a lot more available in the market. The bar for what people spend expert-call time on has gone up, and that's true for private and public markets. What I'll hear a lot is that the stuff we used to spend 2 weeks just getting up to speed on, we do now in a day using expert libraries.

Sometimes, if you're in niche stuff, you still have to go through the cycle because there's not enough out there; it's just a blank spot. But now we're picking 2 to 3 drivers that we really think we need to understand for the investment thesis. Not all of them are best suited to expert calls, but some of those questions are. So then we want to go get 3 to 10 credible experts, really dig into that, and validate it.

So I would just say that spans your direct question, because ultimately the market, the cost of doing this work, and how it's being done have changed. That is true for both public investors and private investors. I'd say the biggest adoption shift we've seen is now a lot of private-market investors, in the last 18 months, mimicking what public investors were already doing 18 months before.

So let me ask you: most people are using expert calls, especially expert libraries. I worry that a running theme of the next few questions is going to be bias—confirmation bias, basically. But I also worry about an echo chamber, right? And I'll give you an example.

You've got growthy tech companies that have the most expert calls in general on Tegus, whether it's the SaaS apocalypse we're in right now, AppLovin, a buzzy IPO coming up, or a few of the kind of cultish tech stocks. I think everybody can figure out the ones I'm talking about or put them in their mind. I worry that you get an echo chamber where you have 1 fund, 5 funds, whatever it is, driving expert-call libraries, and they're coming with bias.

We'll talk about their bias, but if they drive 10 calls on this company in August of 2025, say, and everybody who's looking at the company reads those 10 calls, everybody's thinking about and coming at the company the same way. So my question to you is: do you worry at all about that bias once the library gets published?

I understand there's information outside that, but if everybody's using it, you get biased because everybody reads the same thing. Are funds coming to you and saying, “Hey, how do we think about that bias when we're reading it?” Or do you hear any concerns about that?

Yeah, candidly, we haven't heard that as a concern. I think the other thing to name is that, ultimately, expert insight as a category of insight is absolutely prone to bias. It's a different set of biases that you, as an investor, have to interrogate and apply your lens to.

Obviously, the whole reason why people even use expert calls is we all know management guidance has a bias. Sell-side research has a bias, just inherently, because of the market structure and how that works. Financial data is backward-looking, and so expert insight is ultimately what gives it utility: it is the operator—ideally, the operator's view—to triangulate what's actually true about how this company operates and the drivers and risks that sit in it.

When investors do their expert calls and then those become, like, the top 5 funds are the ones doing the line of questioning around the transcripts you're reading, absolutely, that could be investors driving in bias. But I actually think the bigger bias to interrogate and to be clear-headed about is the bias that can appear in experience.

That doesn't mean that they don't have massive value. It just means you need to be very careful about evaluating what bias this individual might have as you're taking this, and how you think about where to apply what this person is saying. Secondly, I think there's no getting around—and why it's really exciting that the nature of the industry is changing to make this much more possible—the N-count matters.

Still, at the end of the day, the way to avoid bias with operators is to go get multiple operator views. You don't need 30, but relying on 1 operator view to really prove or disprove a thesis is obviously dangerous, right? That's a leap of faith.

You front-ran your bias answer. It front-ran a lot of my questions, both on the expert-call side and when we talk AI, but I'm going to ask them or modify them anyway because I'm very interested in them. Let me again put it into my personal shoes, right? I get on an expert call and talk to an expert. A lot of times, I have a view.

As you said, I generally don't do expert calls as the first expert call, where I've just got no information on the company anymore, right? I've read a little bit. I've got enough to be dangerous. I generally have some bias.

My question for you is: how much do you think experts, when they're on the call, can naturally tell, “Oh, this guy is interested; he's long”? So they're kind of responding to me, my prodding, and being more positive on that.

And how are you hearing other funds think about this? I know when I've gone on a call and I'm bullish on a company and the expert has been bullish, I've been like, “This expert knows what he's talking about.” A lot of times, if I go on a call and the expert's bearish and he can't point to really specific examples, I'm like, “This guy's a clown.”

We'll talk about some expert bias in a second, but how do funds think about their own individual bias when they're coming into these interviews and how it might influence both the interview and their takeaways?

Yeah. Okay. So there are a couple of things. We did a piece, I think it's available online, recently on some of the things that top investors who use expert calls a lot do repeatedly—things they've learned to try to spot and counteract some of these biases that can come on a call.

There are 3 things that jumped out from them. One is that a lot of them do a double-click as soon as they get on to confirm where this person sat in the organization and their purview, so that they understand the perspective they actually had. That's screening through some of that, but it's incredibly important hygiene to say, “This is the lens from which this person is coming, what they saw, and what they couldn't see.” So they've already got that piece right.

Then the second one is, at the end of the day, an expert—someone who's providing expert consultation—is a human, and we know humans are subject to giving you very different answers in the line of questioning when you're trying to go through the same thing. One of the things that a lot of investors will have, they'll say, is their barometer question, which is a way to gut-check this person's positivity or negativity at some point in the conversation.

An example that was given would be, “I'll go through a lot of the questions.”

They'll give me a lot of things about how they're really bullish about the business model. Then they'll throw in a question like, “Talk to me about the culture. How has that shifted?” You can see how a question like that can take someone who's saying, “Hey, all these things are great,” and make them go, “Actually, there's a really deep problem there. Actually, I like—we should speak to that. The culture's gotten a lot worse recently.”

What does that mean? It helps you immediately go, “Oh, well, that's interesting. Tell me more.” So while they might have been very positive on market structure and business model, it starts giving you a hint that there might be misalignment internally. I think that's really unique to expert calls and why they're a very interesting place to find differentiated insight in the market. The more you can treat that as structured, but remember that a human, if you ask open-ended questions and probe in the right way, can unlock really unique insight that's unique to that source of insight in the investment process.

Oh, go ahead.

Oh, go—please continue. Finish.

And then the last thing I'll say: open-ended questions at the end can be pretty revealing. It's really interesting; there's a real parallel with how to interview really well. When you think about interview processes for a candidate you're hiring, they're absolutely prone to bias. Most of the information you're getting is absolutely garbage. Really, it's just track record, verified through multiple references.

One question these investors ask that's also very popular in the way I've interviewed in the past is: “Let's say we're both wrong on what we just discussed, or what we've both agreed to. What do you think we might have missed? What could go wrong?” Those questions at the end are very revealing and sort of go a layer deeper into the things this expert might have missed.

In thinking about risks in the business and drivers, I think one of the biggest things experts are very good at is helping you understand second-order and third-order risks in a business that aren't obvious from the outside.

You know, one of the questions I asked earlier was generalist versus specialist. What I have personally found is, look, if the risk is in a 10-K or something, yes, I can see it. Where I've gotten maybe not the most obvious insights, but a lot of use, is when I hop on a call with an industry specialist and start talking to them. I'll mention something, and then they'll come back with some risk that they live and breathe, that I've literally never thought of, and they'll be able to talk to me about how this specific company is impacted by it.

Let me stick on the bias question for a second. We talked about investor bias—that's what I was talking about. Let's go to expert bias. For me, most of the experts you talk to are one of 2 things. You're looking into Coke, and they're a Pepsi employee, because current Coke employees can't talk about Coke, but maybe a current Pepsi employee can. That's obviously hypothetical. Or current Coke employees can't talk about Coke, but former Coke employees can talk about Coke.

What's the reason most people are former employees? Most people are former Coke employees because there was a round of layoffs, or they wanted to be the CEO and got passed over for the CEO spot, and they left. A lot of the experts I find have a negative bias toward the company. How do you think investors can deal with, address, and calibrate for that negative bias?

Yeah, that's a really fair question. I think the number one thing is just to know that that is a bias. When you're asking questions around risk, you need to understand that they might be overstating what's likely or possible. It's just reality.

The other thing I'd name is that talking to multiple former employees helps you put people on a spectrum, right? If you have 3 out of 3 people saying broadly similar things about the same risk, it's probably credible information. If you have 3 out of 3 all speaking negatively about something, but there are varied levels of tonality in that, then you can make a different assessment.

I think that's how a lot of people have approached that same thing. Invariably, some of these people are going to speak poorly about management or the culture because they left or because of a decision they disagreed with, because they're disgruntled. I think it's—I'm going to go back to interviewing—some of the art of running really good reference calls, which are very similar to doing expert calls, is being able to triangulate where someone is being fact-based in their assessment versus applying a heavy color, heavily colorizing it.

I found that when I conduct references, I have to do 7 to 8 references to really triangulate to the truth. Every time I do that, I get 1 or 2 that, had I taken them at face value, would have really colored the picture very deeply.

Okay, so let me again—and I'm coming at this with my own biases—but let me go back. When you do an expert call, the first thing you're going to do is reach out to your expert recruiter and say, “Hey, I'm looking to do an expert call on Coke. Find me former Coke employees, former Pepsi employees, whatever, who can talk to me about the industry.”

A lot of times, if you're not starting from step 1—you're starting from steps 2, 3, and 4—you're saying, “Hey, I really want to think about how sugar taxes are going to impact Coke, or how ongoing sugar litigation impacts people's view of Coke, or how GLP-1s impact Coke consumption.” So you'll have that.

Then you get experts back. The first and most critical step is picking the right expert, and I find this can be hard. You'll put generally some questions, and experts don't want to answer all your questions, right? They don't want to give the answer away for free, because if they put all their answers in the written question, what's the point of having a discussion?

How can people improve at this screening process for expert calls? How can they get better at choosing experts? How can they ask better questions? And how can they make sure it doesn't suck when they waste time and talk to a bad expert, right? You've generally got to pay them anyway. It's a waste of time. It's a waste of money. How can you get better at making sure you get the right experts?

Yeah. I think the first thing I'd say is that we get thousands of projects every day from investors, and if you talk to a team that services and executes on those projects, they'd say the best outcomes are when the investor takes a hot second to be really specific: “Here's who we want to talk to and why, and the questions we're trying to answer.”

That really helps inform the teams that do this day in and day out. They're able to say, “Okay, well, let me give you some perspective on people that other people have had really good experiences with, that we've already worked with, and then we're going to fresh-source people that we think align with your criteria.”

You'd be surprised at how often people say, “We want to talk to people with this title,” and that's the amount of context. If you do that, then you're not leaning on the teams that do this all day to help you find people who are more likely to fly at the right altitude.

Where this is really common is someone will want someone who can comment on operating leverage, inventory, supply chain things, and they're looking for someone who's just too disconnected from that level of the business and the titles that they're seeking. Seniority is not the same.

On your discrete question there, the answer is we do enough screening questions to see: is this someone credible who can speak specifically to what's being asked, or are they too high-level and unwilling to go there? Ultimately, it's a joint decision. We'll recommend to you that we think this person is credible; we've worked with them before, or, if they're freshly sourced, we're getting signals that this person is faking it and we wouldn't recommend you take them—or they've passed our screen.

This might have applied to some of the stuff we've already talked about, but I do want to hit it again. There are 2 types of calls you can do, and obviously they're broader, but the 2 types of calls in my mind are: I want real-time information. We're not looking for an MNPI. We're not looking for quarters, but you and I, again, we're recording February 9th. There's the SaaS apocalypse.

You might want to talk to someone who's the CIO for a company, and you might want to say, “Hey, how much are you re-evaluating your software budget, your SaaS budget, your per-seat budget right now?” That's a real-time temperature check versus the longer-term question: you want to talk to the CIO and say, “Hey, how are you thinking about Zoom versus Microsoft Teams in the long term?” That's a very specific example.

That's more of a longer-term question, but you might want to look at the overall industry landscape. You might want to say, “Hey, you run Duolingo. How are you guys thinking about the 5-year valuation progression? Where else can you expand Duolingo? You were in learning, now you're in chess. Can you apply it to 4th-grade math? Can you apply it to learning how to play basketball?” I don't know. But that's a longer-term thing versus a more in-the-moment thing.

Where do you think expert calls really excel? Do you think they excel at both? Do you think people see one as better than the other? How do you think people can use these the best?

I think they can do both. I think what's increasingly possible opens up a lot of opportunities that were harder to get to. So I'll speak to both. Generally, as you laid it out, there are deeper questions around understanding business models, drivers, and so on. Those, I think, generally, for a fundamental investor, have been more satisfying in longer, in-depth calls when done properly, because those conversations lend themselves that way.

What you're describing on the former—sort of real-time market impact, what's happening here—is absolutely something. That is a place where people go for real-time insight to get perspective on the market. That's very important, and it will always be there. I think you were alluding to another form, obviously, which is surveys and channel checks. Increasingly, people are treating these conversations as places to collect signals on trends and specific data points, and that is where more and more people, unless they have really sophisticated internal setups to do that, have found experts frustrating or unreliable.

What has changed? First, they were just incredibly cost-prohibitive. The cost to operationalize those for an expert network didn't look that different from an individual expert call. You're not going to spend $100,000 for a single survey, whereas you could spend $2,000 on an expert call. But AI is actually one of the biggest areas where we're early, and I expect it to have a big impact on your question of where expert calls are going to be most powerful. I think that for things like survey and channel-check insights, AI makes the entire cost model and operating model behind that vastly different from what we've ever experienced in the industry.

With AI interviewers, they're not human and don't have to arrange a time. You could have them go talk to 10 CIOs on their clock, based on their availability, and get really quick insight on a question like that in real time. That was just stuff that was really hard to operationalize even 6 months ago.

It's so funny, because the way I've structured this interview and my notes is expert calls in the front half and AI in the second half. This is like the fifth point where we've hit the end state, and I'm thinking, "I should talk about how AI is going to evolve this thing." Even just doing this interview, you can see how AI is creeping into a lot of these things.

Let me ask about note-taking. I just did an expert call last week. I think you and I did a prescreening call on Wednesday, and I was literally coming from an expert call. I do an expert call and read an expert call, whatever it is.

One of the tough things I personally find is keeping track of note-taking on these expert calls. I'll highlight things in the Tegus or AlphaSense app, and I'll write down notes, but it can be hard. You read 4 expert interviews over 6 months on Company XYZ, and it can be hard to remember these things. It's hard to remember anything you read about a company, but especially an expert call, it can all blend together.

AI, when we get there, will probably help a little bit. But how do you find the best people, especially in real time when they're doing the interviews? How are they taking notes? What are they focusing on so that they remember and ingrain whatever learnings they're getting from these expert calls?

I think a best practice is obviously to book enough time right afterward to synthesize and take stock. But I wish that skill set and that discipline were already obsolete for us.

All road maps are leading in this direction. You do an expert call through Tegus, and it's table stakes that it should be recorded, instantly transcribed, and sent to you, which we do today. More importantly, there should be an AI summary and synthesis that mirrors the way you want to organize your note-taking around it. The fact that we're not there yet—I think within months, most people are going to be moving in that direction.

Traditionally, the funds that have done this really well and systematically have had a discipline around taking the notes as soon as the call is done. Those notes go into an internal drive that everyone can extract insights from. The other thing I'll say, which is a really big part of the next conversation, is that traditionally people have thought of expert calls, all these services for proprietary research and investment research, traditional data feeds and other providers, AI tools, and internal content as separate things.

Increasingly, what's happening is that you're doing expert calls as a firm all the time, you have investment memos, and then there are external data providers. Plugging all that in and using AI to extract those insights is ultimately where things are going. When we get to the AI conversation, I'll talk through some use cases I'm seeing that are really interesting and how insights are coming out of that.

Ultimately, I think the world of having to be a really expert note-taker after your call has a very short half-life. One of the things AI should be able to do for you is make that not a huge part of your routine. You should be able to have the technology immediately send you a summary of exactly the insights and structure you want. The technology can do that.

We're going to keep coming back to AI, so let's start transitioning to it. In my head, the AI discussion has almost 2 parts. There's using AI tools in general, and then, because we started with expert calls, there's how AI tools are shaping and evolving expert calls. That's obviously a subset of the broader discussion, but I think it fits here.

Let me start with the same question I asked about expert calls. If I'm a listener, whether I'm using AI on expert calls or using AI in general, and I'm going to walk away from this conversation with 1 thing about how I can use AI to be a better investor, how would you answer that?

I'll tell you where we're seeing all the action for public-markets-focused investors. One use case where you can immediately start getting leverage and making your life better is around earnings. The number-one thing you need to do is pick a place where you find yourself spending a huge amount of time doing hand-to-hand comparison and synthesis, taking multiple data sources, and forming a view under time pressure. That is ultimately where AI is strongest, and earnings season is where we're seeing that in public markets quite a bit.

I'll give you some examples. There are things people habitually would have to do when they have a name in their portfolio. This is a real investor conversation: "I'm looking at Reddit. They just published earnings, and management guidance was very positive. Now I've got to basically update the thesis on whether or not we want to stay in the stock and what's happening around us." The things you used to have to do very manually, you can now do within hours.

One of the prompts this individual has set up is: "Here's management guidance. I want you to compare what the CEO is saying to the actual cash-flow statements of the last 5 comps that I tell you have already reported." What that's allowing people to do very quickly is say, "This individual is speaking positively, but the cash-flow statements show that there's a lot of negativity among the others. So what does that tell you?"

There are 2 possibilities. 1, Reddit is an outlier and things are going really positively. Why? Or 2, management is overconfident, and we're already setting up for a question mark there. These are the types of things that are happening around earnings.

What AI is really good at is synthesizing insight from multiple sources and drawing connections that are very hard for a human to make quickly. That's probably the number-one place I would point public-markets investors. There are multiple things that people are doing right now, all the time.

That's super interesting. But if I could push back on you slightly.

Sure.

This is Yet Another Value Podcast. On my average podcast, a guest comes on and we talk about 1 stock for an hour. It's a deeply researched, generally concentrated investor. When you say "earnings" and things that need to be done quickly, my first thought is that you're talking to pod shops that are trading quarters and whisper numbers and all that sort of stuff.

So let me reframe the question. If I were ignoring immediate-term considerations, how would someone who's a 5-stock, concentrated, long-term investor use AI to evolve their process?

I think there's another area. When you're going to take a position in a company, there's ultimately a heavy, heavy amount of work involved in understanding the fundamental drivers of the business and determining whether you can get a differentiated view versus consensus.

Yes.

I love that you said “differentiated view” there. Yep.

Yeah. And I think ultimately, some of the really interesting use cases there are around how consensus is formed across multiple layers. What are sell-side analysts saying about it if it’s a widely covered name? What are the key debates on the sell side, and what are they saying about it? What are all the people saying? What are all the experts saying about the key drivers that matter? And then what is our internal view on those? You can triangulate and compare those perspectives.

One thing that I think you had asked me coming into this is: What is AI uniquely really good at that surpasses the ability of the average investor, versus where it’s merely coming up to the ability of what a junior analyst you bring into the fund can do? One thing I will say is that it has the ability to synthesize and compare perspectives across tons of different sources in a grid-like format.

One of the things that I think we’ve seen investors using more in fundamental research is that you can look at so many different components and compare them. What is management guidance saying on this? What is the sell side saying on this? What are the expert calls we’re doing saying? How do they compare to what’s being said? What are the expert calls in the transcript library saying?

I think that’s allowing people to say, “Hey, these are the real debates on this name that are fundamental to the value-creation story, and that’s where we’re going to do a lot more work.” That’s the kind of stuff that you just wouldn’t know to do at the level I’m talking about. An 80-by-80 grid comparison of inputs across multiple data sources is just not feasible for a human analyst to do. But that reveals really insightful places for investors to go and dig deeper.

We’ll probably come back to this, but one thing that just jumps out to me is that there are some names on Tegus where there are 80 expert calls a year, right? There’s no—I mean, maybe, but if you’re saying, “Hey, I’m going to follow 30 companies,” there’s no effing chance you’re going to read 80 expert calls on 30 different companies. Yeah.

AI can do it in half a second and summarize it for you, right?

So I want to ask 2 questions on that. The first question—I know I’m not alone in this—is that there are lots of tools that will automatically build financial models for you and extrapolate them. You know, Comcast reports Q3 earnings; they’ll automatically put it in, update the model, everything.

I build all my models by hand, especially as I get close to making an investment, because there’s something about going and doing it that makes me learn, makes me think, and all that sort of stuff. Whereas, if I just had it presented to me with AI tools, I kind of worry about that. If I just had AI summarize 80 expert calls for me—now, 80 is a lot, and going and reading them all is a lot—there is something about getting the summary that maybe I don’t quite understand or internalize as much.

So when you talk to firms, especially portfolio-manager-level people, how are they talking about that trade-off? On the one hand, I could never read 80 expert interviews, especially across 30 names. On the other hand, if I just get 80 summarized for me, I don’t internalize it or think it through as much. I’m losing that edge, that insight; I’m just outsourcing it all to AI. How are you hearing people talk about that trade-off?

Yeah, look, I think it’s a fair trade-off, and it’s a very understandable emotional reaction. I mean, I’ve had it myself. I went through the experience of building company models, and I know that what you’re describing—clicking through the drivers and the sensitivities by actually building the drivers myself, and running the sensitivities and the scenarios through it—there’s real value in that.

Here’s what I’d say, though: I believe in my bones that by 2030, there are going to be really high-performing portfolio managers in this next generation coming up who absolutely never had to go through that. They’ve never built a super-detailed M&A model, and yet they’re pretty good at leveraging this stuff to get to insights, triangulate on what really matters, and get good investment outcomes.

The debates we’re having in the industry are more about exactly what you said: Until I can fully trust this stuff, it’s still prone to errors in judgment and data that I just don’t trust or believe in. I think a huge part of our philosophy in how we’ve built AlphaSense is that everything in AlphaSense is fully traceable down to the source.

That’s really important because when I go through workflows, even for my own research for go-to-market, I need to see instantly where that insight is coming from. Otherwise, it just interrupts my workflow. I don’t want to get 2 hours in and then suddenly have it all be on a shaky foundation.

The second thing I’ll say is that AI is very prone to—if you prompt it a certain way, it’ll pound the table. I had that experience where I said, “Build my go-to-market plan for AlphaSense through the lens of a CRO reporting to a board.” The conviction it will give me on certain things makes me go, “That makes no sense.” My judgment suggests that, while that might be true, there was a verbatim series of calls we had with customers saying X was true. I know enough that the TAM of that segment doesn’t make any sense for that recommendation.

Ultimately, for the investor, I’d say the value that comes from judgment and understanding market structure and business models goes higher. But for those of us in the industry—I left the industry, but for those who stayed—a lot of your sense of self as an investor is your technical prowess and analytical skills. I think those are getting commoditized over time, and what’s much more important is your pattern recognition, judgment, and ability to push on these things.

Look, everything you just said, especially toward the end, matches my worldview. So let me ask this: You mentioned—if I’m quoting—having a differentiated view when you’re making an investment, right? That’s kind of what you’re looking for when you’re making especially a concentrated, long-term investment.

If everyone is using AlphaSense and AI to summarize the same AlphaSense expert library—this is why I don’t read sell-side reports, right? If you read all the sell-side reports and then make your conclusions based on that, you’ve kind of just got the market view, or you’ve got the sell-side view.

If everyone’s using AI to summarize everything, how are people thinking about, “Hey, that’s the table stakes, right? I need that. I need that basic information. How do I get a differentiated viewpoint? Where is my special sauce, where I’m going to have a differentiated viewpoint, when everyone else is using the same AI to summarize the same expert calls?”

Yeah, I think with a lot of these technology innovations, it just shifts the baseline. You can think of doing financial analysis before the PC and Excel, right? Having these really sophisticated ways of doing that was no longer a differentiator; that became the baseline. If you weren’t doing financial analysis that way, you were behind.

I think where we’re getting to is that it’s always been about access to information and your ability to have an investment process that yields results others can’t get to. We’ve always talked about how markets are efficient and everyone has access, but we know that’s not true. That’s why we were all trained to sweat the notes, go deep into the 10-Ks and 10-Qs, really synthesize all these disparate things, and get to something differentiated, even before we talked about getting an edge through alternative datasets.

What’s happened with AI is that the technology is so powerful that any gains from it are getting harder to come by. The alpha comes from some of the same things we’ve always talked about: the ability to have these systems work for you so that you can make decisions much faster, with conviction.

I’ll give you an example in private markets. I’ve seen this really come into play in the last 12 months. There are very big parallels to a long-term, concentrated public-market investor—there are parallels to a private-equity fund that makes a couple of concentrated bets a year.

Yep.

Yes. And when I’ve asked them, “Hey, how’s this impacting you? Are you looking at more names, more opportunities?” the answer is yes. “Are you making more investments per year?” No. That’s not our strategy. We’re still only going to make 3 to 5.

But we are much, much more convicted about those 3 to 5 as a result of what’s possible. The due diligence we used to do to get from point A to point C in our investment process—the time to get from point A to point B in our process has compressed to a day from weeks. Therefore, the amount of time and energy we spend on diligence in points B and C has increased, which is usually where the key debates in the investment committee happen: where the value creation comes from, what the drivers of the business are, and our differentiated view on those.

That’s where all the real work is going.

Do you think they should be going? So, you said 3 to 5, and they say, “Hey, we’re more convicted.” I think you suggested at the beginning that because they can go from point A to point B faster, they should perhaps be doing 8 to 10 instead of 3 to 5.

Should it be the other way? If they’re getting more convicted and they’re able to go deeper into point B to point C, which is probably where they’re addressing the real niche cases and their real differentiation, instead of 3 to 5, should it be, “Hey, we’re more convicted, so we should be more concentrated. We should be doing 1 to 3 instead of 3 to 5”? Do you think that should be the right answer?

Yeah, I don’t know, because I do think there are some funds that have said, “Yeah, it actually has increased the amount of things we’ll do in a year,” right? And there are others who are saying that’s just not our operating philosophy, and we’ll only do the 3 to 5 that we usually do. And, yeah, sure, maybe some have been like, “We have even higher conviction now, so we’re going to bet the fund on 1 or 2 ideas.” I haven’t seen that as much.

I think the general principle, though, is that everyone recognizes valuations are elevated. It’s more competitive, and there’s more to put to work. So when we bid for these good assets, we have to be much more convicted. That’s the scarcity. Therefore, so much more of the work is making sure that we have a credible story for how we’re going to have value creation and a real exit.

That bar has just shifted dramatically over the last 2 years, not because we chose it to, but because we can feel it around us—how quickly people are moving on opportunities with conviction. We have to stay in line. I think that’s ultimately what’s happening.

No, I just asked because that’s exactly what you’re saying. I have some friends who used to do, let’s say, 5 investments per year, and now they’re like, “Hey, because of the AI tools, I can get to these faster, so I do 10.” And then I have friends who say, “I did 5, but now I do 5 with a lot more conviction.”

But I haven’t had anyone say, “Because I have more conviction, I do 3 instead of 5.” So I was just—I haven’t heard that yet.

Yeah. But I do want to underscore, too, though, that while the end result in the funnel might still be the same 3 to 5, the amount of things that get looked at before they even get to that has expanded. I think that’s ultimately the point. You think of how many assets you can look at that might get there if that universe has expanded, sometimes materially.

Some have said, “I’ve looked at twice as many things now,” because you get a CIM, you can analyze that CIM instantly with AI and all of our internal stuff, and get a green, yellow, or red in a way that took weeks of analyst capacity. So I think that’s been a huge difference.

Yeah. Earlier, you were talking about how, 50 years ago, financial analysis was literally Excel spreadsheets. Before you put it into a computer, there was literally a physical piece of paper—a spreadsheet—that you would build everything out on, right? So eventually that goes online, and that gets commoditized. Now there’s all sorts of stuff that will automatically build off Excel.

So I would posit to you that 60 years ago, you could make money with quants in your head if you were a really good, literal financial analyst, right? You could make money by modeling. Think about Ben Graham just calculating net working capital.

Totally.

I would posit that maybe 10 to 15 years ago, you could make a lot of money, probably more so on the qualitative side, right? The financial analysis got commoditized. The qualitative is where all the money was made.

And I would just say, look at the past 15 years. If you bought—or if you thought through—Google, Facebook, or Amazon, whichever one, these are the best businesses ever. The world is trending that way: the internet, increasing returns to capital, scale, all this sort of stuff. If you could figure that out, that was not a spreadsheet number. That was qualitative. That got you there.

AI is kind of—I’m not saying it’s replacing the qualitative, but AI has really raised the bar on qualitative. What do you think the next skills are that generate alpha? If financial analysis has already come down and a lot of the qualitative comes down, there have to be some skills that get elevated.

Whereas 60 years ago, if you were great at the qualitative and terrible at the financial, you couldn’t make it work. But then when the financial gets commoditized, qualitative makes it work. If that’s coming down, what’s the next skill set, do you think?

Yeah, I’ll give you my thesis, and there’s a lot to be proven out here. I’ll talk about private markets first, and I’ll talk about public markets, because I think there are some parallels, but they’re going to be different.

I think on the private-market side, what’s been happening is that the returns from being really good at financial structure and dealmaking have been going away. I think that’s widely discussed in the industry. So what AI will probably help with is facilitating the ability to act really quickly on a much bigger opportunity set and win more deals when they fit in your strategy.

I think firms that focus on portfolio value creation post-close have a lot of opportunity. We didn’t go there here—this was more of a lens on AI and the investment process—but I think one of the other things is that I was at a mid-market PE conference last year, and all the buzz in the room was about the things people could do by taking AI to portfolio companies to drive value-creation stories.

So I think that’s a likely place where some of the big gains will come from: using AI really effectively to create more places to look and get higher conviction on deals in the way we discussed. But I think a lot of it will also translate to portfolio value creation, because AI has a real, fundamental set of use cases where it’s changing operating models and cost structures that make sense in that world.

On the private front—this is hard on the public side, which is where I’m focused—but on the private front, I actually think it’s going to be this: if AI is a tool that everyone can use, you have the old Warren Buffett idea that if you’re at a parade and you stand on your tiptoes, you get a better view, but then everyone stands on their tiptoes, so no one’s better off. Actually, everyone’s a little bit worse off.

I actually think financial analysis gets commoditized by AI, and a lot of the qualitative gets commoditized. I think the people who are best at human resources and people are actually going to be the people who—I think that’s going to be a skill set that gets elevated on the private side.

But I don’t smoke, so maybe I’m just smoking something, or I’m just too far out there—galaxy brain. What about on the public side? What do you think skill sets get elevated?

Yeah, I think this is the common discussion in every industry, which is that there’s a long-term problem with this answer. But I do think one thing we’ve talked about is this shift from the analyst skill set to the architect skill set: people who are really adept at using these things to create leverage in the investment process.

For a concentrated 3- to 5-name, long-term investor, this is probably less resonant, but I do think this will impact public markets. I think you’ll see a lot more people using AI in fundamentally focused investment work to do portfolio monitoring, idea generation, and to look at a lot more things a lot more quickly.

I think that’s going to change the stuff you said, like how pod shops behave. I actually think that pressure is going to move into more places in the industry. And then, long term, I think the real question mark is what happens to fundamental investing in the way you described.

Ultimately, do we have this cohort of people who grew up in the world that you and I grew up in, who are deep experts in it and understand it through years of investing and pattern recognition, and do we lose that with another group? Or does this new group that comes in leapfrog that somehow and start looking at truisms that we’ve all lived with that are uncorrelated in the data and actually don’t matter?

And then there’s a whole different version. It’s not quant investing, it’s not fundamental, but it’s something in between, right?

Yeah, I’m very worried. I’m already a dinosaur. Let me go back to our bias discussion, right? This is something I think about a lot.

But before we go there, just on the public-market side, I have to ask for my own curiosity: on the portfolio-monitoring side, how are you seeing concentrated fundamental investors use AI for portfolio monitoring?

Yeah, I think really big examples would be—there are extreme scenarios, and then there are day-to-day scenarios. The extreme scenario was Liberation Day last year, right? We saw people who had these portfolios and were instantly asking, “What is my exposure, and what are the recommendations? Where should I go dig across 10 or 15 names?”

What AI was very good at was, in those kinds of fire-drill moments, indicating within hours all the places—all the different research that was different from their view and aligned to their view, and where their exposure was. And then that’s where the work was done.

I think that is an extreme example, but we also saw that again, actually, as you described, more recently, around all this bearishness around SaaS and AI exposure. People have been using AI to very quickly get their head around things like that.

From an ongoing portfolio-monitoring perspective, ultimately what really matters, though, for this to work well is that it’s only as good as the number of data sets you have access to in the market. But ultimately, I think the market has shifted from things that help me go find answers to questions I’m looking for, to things that help me produce these workflows I’m constantly doing, to now custom autonomous things that run reports as if I had an analyst working on it.

So people are using portfolio monitoring to say, “Every Friday, I want a report in this format that tells me trends and inflections on these parameters against my portfolio.” And those are the types of things where portfolio monitoring is just an always-on, custom way. Just imagine if you had infinite analyst resources: What would be some nice-to-have discretionary things you’d ask for that would make you feel more in command of your portfolio? That’s the kind of stuff that AI does pretty well.

Right. Let me go to bias real quick. There are 3 types of bias I could see in AI, right? If I’m crafting prompts for AI, there’s bias in myself. If I’m crafting a prompt on a company I’m bullish on, I can bias myself in the prompt.

There’s bias on the company side, right? If I have AI read every investor day and every earnings call a company’s ever done, management teams are generally pretty darn bullish on themselves, and they’ve got a lot of bias in the way they present. And analysts aren’t exactly going to get on and scream at the company, because then they’ll get cut off and they’ll never get to talk to the company again.

So I worry about bias for myself when I ask. I worry about bias on the company side if I have AI trained on a company’s data set. And then, on the expert side, if I have AI read a bunch of expert calls, as we talked about with expert calls, experts, in my opinion, tend to be a little bit more negatively biased. So if I have the AI train on expert calls, I worry about negative bias in the training data. How are investors thinking about those 3 biases when they’re using an AI tool?

Yeah, I think this goes back probably to the last question of where the skill of an investor becomes differentiated over time. And I actually think, look, bias is inherent in almost every data set you can look at to evaluate an investment.

What’s different now is you can triangulate multiple of these sources, with their biases, in a way that was really hard to do as comprehensively. And so I think what investors are doing is, rather than trying to oversolve for how to eliminate all bias, there’s a recognition that they all have that, and they’re using increasingly sophisticated ways of comparing and contrasting sentiment and perspective to see where the debates are, and then forming their own independent view: Who’s wrong and who’s right? Management obviously has a certain bias, and as you said, these experts might be really negative. But where do we think the truth lies, and how can we get smarter on that based on what we’re looking at here?

I don’t know that that was a super-direct answer to the question, but I think that’s just what I see happening. In a prior world, you had fewer sources you could evaluate in the time you had, and they had bias. Now you have more sources you can evaluate, all of whom are biased, but the triangulation you can do across these different sources and perspectives is infinitely higher than you could before. And that’s ultimately, I think, great investment work.

There’s no perfect answer, I hear. Let me end by asking about AI and expert calls. AI really shifts the use case, especially for expert-call libraries, but also for expert calls. There are 2 ways that we’ve kind of hit on, which I’ll just summarize.

Number 1: If I wanted to do 100 expert calls on a company because I wanted to dive really deep, I can’t; I’m limited by my own time. I could have an AI agent serve as the questioner, ask 100 things, and do that. Theoretically, I could have that happen, right? That’s number 1. Number 2, I can’t read 80 call transcripts on 80 different companies. AI can. So there are 2 ways that fundamentally now you can use more expert-call-library transcripts, and maybe you can get more expert calls if you want to. How are you seeing AI and expert calls evolve together? How are you seeing your customers who are at the far, far tail end of using AI and expert calls? How are they marrying the 2?

Yeah, I think, ultimately, expert calls have always, as we said in the first part of our conversation, been a really unique source of insight because they’re human and varied, and they’re not going to give you yes-or-no answers, right? You can tease out a lot from them.

When we talked at Tegus, as we were building the business around expert calls, we were saying this is probably one of the most unique data assets in the market. You can think about the amount of expert knowledge that sits out there on all the investable markets, research names, and companies, and it’s off-platform. It’s not—you can’t extract it anywhere.

We had these business-model innovations that opened up the number of expert calls that could be done and how much could be captured and searched. That was version 1.0 of the Tegus model versus the traditional model. Then, 2, what AI is basically doing is pushing that trend further. If the next gate was investor time to actually conduct those calls, that’s no longer a constraint, right? So ultimately, it’s just the amount of resourcing available to go run at all these things.

To answer your question a little bit abstractly, I think one thing that’s really interesting is the amount of expert insight out there that can be captured, queried, looked at longitudinally, and compared and contrasted over time. That is a real-time data asset that’s building every day. And that wasn’t true a few years ago.

Some investors, especially really large ones, are recognizing that they do massive amounts of expert calls themselves. Some of them are crossover funds across public and private markets, and they’re comparing insights against those. So now you suddenly have 2 different data sets: what’s happening in private markets that we can see and what’s happening in public markets. Does that help shape our conviction on different names?

I think that’s where AI is an accelerant of a trend that was already happening in the market around expert networks. I think investors are really seeing this as one of the more unique data assets that is being built, and they want a stake in it. They also bring their own proprietary stuff that others can’t see to it.

One of the biggest things we’re seeing is that a lot of investors initially were just looking for an expert transcript library and AI tooling to search it. Increasingly, they’re also bringing their internal content alongside it. This is much more relevant, I think, for larger, well-resourced funds that do a ton of work, but it’s a big trend in the market: They’re able to see things that others can’t because of all the research they’re doing in the market across disparate teams.

Softballish question, and then I have 2 more questions. Softballish question, and then I’m going to end with a true knuckleball question. We focus on investors, both public and private. AlphaSense does a lot with companies. Are companies going into the expert-call library and using expert calls to source ideas, think, and change strategy, or even just seeing the questions investors are asking to change how they’re responding to IR?

Or you could also tell me, “Dude, the companies are the experts. They don’t need to go to an expert library. They can just call up their supply manager and have them answer.” So I’m curious if you’re seeing companies adjust and adapt to how both expert-call and AI libraries work.

Yeah, look, companies are a big part of our business. At AlphaSense, almost 40% of our business is built on corp dev, corp strat, and IR teams. And so they’re huge consumers of the same insight that investors look at. Their use cases are nuanced and slightly different, but I think this went from an industry that was very focused on fundamental investors to now becoming very much a core part of how sophisticated corporate decision-making is made.

Yeah, I was just wondering exactly that—whether they’re using it or not.

Okay, knuckleball question. Super weird, but if I can give you the background, in 2011–2014, there was this big Chinese reverse-merger fraud in the stock market, right? You would read the 20-Fs of these companies and they would say, “Hey, you know, we have 600 million acres of woodland in China.” And people would say, “Oh, well, an acre of woodland’s worth a dollar. 600 million—this is worth $600 million. It’s a buy.” Well, it turns out 600 million acres of woodland doesn’t even exist in China. All these things were frauds.

I wonder if there is a return to—if the scale and returns to fraud improve in an AI age—because, you know, if you’re a company and you’re running a fraud and you’re getting the 10-K and AI is just detecting it, and they don’t have that human who’s going and saying, “Dude, their headquarters is like a PO box in Boca Raton.”

Now, yes, it's the front page of the 10-K, but the human person who goes and says, “This management team is out of their mind.” Do you think AI increases the return to fraud, or the return to far-left, really nasty companies, because if they just get bucketed into this big quantitative AI pool, it's tougher for them to detect? Does that make sense?

Yeah. Could you say that one more time or reframe it just slightly for me?

So I'm just wondering—if I'm thinking about the Chinese reverse-merger frauds, that's really what I'm thinking about, right? If I went to AlphaSense and said, “Hey, find me undervalued companies on an asset-value basis,” and it was just reading the Chinese reverse-merger fraud 20-F, it would say, “This is the best value buy in the stock market,” right?

Every other peer with woodland acres trades for $1 per acre. This is trading for $0.05 per acre, and it would be telling me to buy, right? There were a lot of these things out there. I just used the woodland example, but there were a lot of these things out there, and it took somebody calling around, going snooping. Plenty of investors fell for these things.

But I wonder if, in 4 years, all these things that a human reading them would say, “Hey, there's something wrong here,” or a human who literally flies out and says, “Oh, this $4 billion company has its headquarters on the third floor of a mall. This is kind of weird”—an AI wouldn't see that. So I'm wondering if AI increases the returns to fraud because, as you get more quantitative money and more quantitative processes, that kind of human check goes away.

Yeah. So, 2 thoughts on that interesting question. I think the first thing that comes to mind is that, ironically, AI is being used a lot in the fraud-detection industry, right, to find patterns and things that indicate that something is amiss. I'd say that I don't think there's anything inherent about AI that suggests it becomes an accelerant for the fraud that's possible, because I think it can equally be, when used right, a pretty powerful weapon for detecting fraud in pretty idiosyncratic and straightforward ways in other industries.

I don't think there's anything that stops it from looking at, to your example, visual imagery—satellite imagery—and saying, “Hey, there is something that's mismatched versus company guidance.” We can see from imagery on shipping lanes that the traffic is not even remotely what the company guidance is. I actually think it could be powerful in helping investors parse those pieces that used to require having someone on the ground or sending someone to go look.

On the other hand, what I will say—and this goes back to the core of what we're discussing—is that AI ultimately does some things in a superhuman way. I think, ultimately, that is synthesis and finding really discrete details and connection points in a way that humans cannot replicate—what AI is able to do in that domain. On the other hand, it is absolutely not at the level of a PM or sophisticated investor on anything that remotely looks like an investment recommendation, right?

So think of AI as—I think what I get really excited and bullish about is that I love underdogs. I ultimately think what AI has enabled here is that it has absolutely collapsed the resource advantage that the biggest funds have had versus mid-market funds. You can basically go do stuff as if you had a crack team of incoming KKR analysts, right? It can do a lot of the stuff they're able to do.

But what it can't do is what very likely you can do, which is look at that report and have your spidey sense go, “This doesn't make sense. I have to go dig deeper,” right?

No, it's funny you say “level the playing field,” because I do worry. As a small investor, you have a lot more nimbleness, but you mentioned it when you were talking about AI use cases: the big funds with lots of off-market data. I worry that there will be no more role for a small investor because AI levels the playing field so much that it's the larger funds that are sending their analysts to every industry conference out there, having them put notes from every industry conference, and getting all this data and analytics that a small investor just can't do.

I worry that the returns to scale actually accrue, and there's kind of only a place for larger funds that are generating literally proprietary information by sending people in person to do all of this different stuff. But that's probably a conversation for another day.

Ryan, this has been so much fun. As you can tell, I think about this stuff and where it's going all the time, and you were the perfect person to have on. Any last thoughts you want to add—expert calls, AI, anything? I think we've been pretty comprehensive, but I could probably go for another 2 hours, to be honest with you.

Yeah. No, look, I really enjoyed the discussion, and I think the number-one takeaway I have for your users is that I think there's absolutely going to be a role for fundamental investing. I think the thing—

I got my fingers crossed—like, absolutely. And I think these debates that we've talked about, like whether being a generalist or a specialist, I don't think that there will be wildly successful specialists and generalists in the AI future. I don't think that changes at all.

I think some big funds will have absolute advantages from what we're talking about, but then I think there are going to be a lot of smaller funds that are really nimble with this stuff that outcompete them. I don't think that story changes. I just think, ultimately, like all major technology changes, the baseline for what's possible will shift, and people will have to figure out very quickly what's table stakes to not fall behind and what's a true advantage.

I think we've talked a bit about what that looks like in practice right now. There's a lot of hype, but there's also a lot of real stuff happening. So that's—

Perfect. Well, Ryan, hey, look, I really appreciate you coming on. Again, these are just things I think about all the time, and I appreciate you walking me through them and helping me get a little bit better at using AI and thinking about how to use expert calls. So, Ryan Fennerty, AlphaSense, thanks so much.

Great. Thanks, Andrew.

A quick disclaimer, nothing on this podcast should be considered investment advice. Guests or the hosts may have positions in any of the stocks mentioned during this podcast. Please do your own work and consult a financial adviser. Thanks.