Will This $11B AI Startup Disrupt Big Law?

Ed Elson speaks with Gabe Pereyra, President and co-founder of Harvey, a legal AI startup. They break down what AI can realistically automate in law, how quickly that shift could happen, and what’s likely to slow adoption. They also discuss how Harvey has built trust with major firms despite being a young company. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:04Speaker 00

  What's driving the markets this week? What's on investors' minds as they look ahead? Find out on the Markets Podcast from Goldman Sachs. A breakdown of market moves and macro signals in 10 minutes or less. The Markets Podcast from Goldman Sachs. Listen now.

• 0:31Speaker 01

  Are you a pet owner? Every six seconds, a pet owner in the U.S. gets hit with a vet bill over $1,000. Always an unwelcome surprise. That's where Fetch comes in. Fetch is the most complete pet insurance for dogs and cats, according to consumeradvocate.org. You get paid back up to 90% of vet bills at any vet in the U.S. and Canada, with claims paid back in as little as two days. Go to fetchpet.com/save right now for your free quote. That's fetchpet.com/save.

• 1:10Speaker 02

  Your heart can tell you a lot about your health. Apple Watch Series 12 measures your heart rate every five seconds with the most accurate heart rate sensing in wearable. So your vitals app, now with heart rate variability, can tell you when something is off. And your readiness score can let you know when to rest and when to push. Hear the story in every heartbeat with Apple Watch Series 12. The features described are for wellness purposes only and not for medical use. iPhone 11 or later required. Based on Apple conducted study of heart rate accuracy, August 2026. Visit apple.com/aplewatchseries12.

• 1:46Ed Elson?

  Welcome to the Prof G Markets Founder Series. I'm Ed Elson. Since the start of 2026, investors have been asking one big question: How much of the economy will AI disrupt? We've already seen the fallout in software, where waves of SaaS apocalypse sell-offs wiped out hundreds of billions in market value, but that may just be the beginning. Now, attention is shifting to the legal industry, a trillion-dollar market built on manual, time-intensive work like document review, due diligence, and compliance, exactly the kind of workflows that AI was built to transform. Well, my next guest saw that as an opportunity.

• 2:27Ed Elson?

  In 2022, he built an AI company designed to streamline legal work at scale. Now, the question is, can it actually disrupt one of the world's most archaic industries? Well, with over $1 billion raised, an $11 billion valuation and adoption in over 60 countries, it's well on its way to doing just that. This is my conversation with Gabe Pereyra, co-founder and president of Harvey. Gabe, welcome to the show. Thank you so much for joining me. So much to get into here. I guess maybe lay out for listeners, what actually is Harvey? And how did this company get started?

• 3:07Speaker 04

  Thanks so much for having me. I would think of the problem that we're solving with Harvey is how do we help large law firms and their clients, which are large enterprises, and increasingly all law firms and all companies, go through exactly the transition that you talked about? So when we started the company four years ago, we built, I think, what most companies built of some form of co-pilot for a professional. Cursor and cognition built this for programming. We built this for legal. And I think what you're starting to see the shift as these models get better and better is you need to start thinking not just about the productivity of individuals, but the productivity of entire organizations.

• 3:50Speaker 04

  And what is the infrastructure that they need to be able for the entire firm to operate effectively? And so that's a lot of what we're building at. At Harvey.

• 4:00Ed Elson?

  When did you come up with this? And how did you realize that this was going to be a huge opportunity?

• 4:05Speaker 04

  Yeah, we started the company summer of two thousand twenty two. I was I was at Meta on their large language model team. G P three had just come out. I had been doing A I research for about ten years before that. And so even in two thousand fourteen, I think I was at Google Brain and DeepMind. A lot of people in that community had the belief that we would figure out a way to build these systems. I don't think the path was super clear. I think we're kind of still being surprised by the way it's going. But there was just like I had this strong belief of we will be able to build superintelligence, A G I, things like this. And I think when you do a lot of that research, you're constantly thinking about what problems are the models good at? What problems are the models not good at?

• 4:50Speaker 04

  And in two thousand twenty two, my roommate was Winston, who's now the C E O of of Harvey. He was a lawyer at O'Malvany. And I'd been brainstorming startup ideas as I saw these language models get better. And one day he kind of showed me the work he was doing and the workflows. And that was kind of the lightbulb moment where at the time, G P three couldn't do that work. But it was very clear they would keep getting better. And that felt like one of the big industries that would just be a very clear application of this technology.

• 5:18Ed Elson?

  It does seem as though legal work is basically ground zero for A I. Or at least that seems to be the way people view it at this point. I mean, what we've seen is that A I appears to be able to do all of the grunt work of the white collar jobs. That is kind of like the the starting point of what A I can do. And it does seem as though law is exactly that. Just for those who maybe don't understand how law firms work and and and what kinds of work they're actually doing. Could you tell us a little bit about the legal industry? What are the workflows of a lawyer? What kinds of things could A I potentially automate in in that business?

• 6:01Speaker 04

  I think when most people think of law, they think of consumer law. And so I need to review a lease or, you know, I need kind of look at one document. And the models are obviously great at that. And I think to a large extent, the base models can do that. And then there's kind of corporate legal work and particularly big law, which is the massive law firms. And you can think of the work they're doing is the highly specialized legal work where you need these incredibly talented, very specialized partners. And I think kind of the two best examples of this are you're doing a massive merger, right? Or an acquisition. You want to go buy a company for $10 billion, $50 billion. Like, you need the highest tier partner to advise you how to structure that transaction, or

• 6:44Speaker 04

  you're doing about the company litigation, right? Like a big antitrust or something like this. And the way that I would think of the workflows of all these firms is these projects take thousands, they can take tens of thousands of hours, teams of associates. And the big challenge is, for example, when you're going to buy a company, you need to go understand all of the contracts in that company, all of the things that are going to happen when those contracts change because of the merger or the acquisition, all of the legislation around it. And then there is also all of the negotiation dynamics, right? It's a semi-adversarial or it could be an adversarial process. Same on litigation. And so to your earlier point

• 7:24Speaker 04

  of, I think when language models came out, legal was kind of this great application where typically your workflows as an associate is you're getting emails from senior associates or partners, and they're just giving you tons of tasks. Go research this. How do I write risk factors for this document? And what these associates are incredible at is they can just absorb these tasks. They know how to go use all these tools and solve all these problems. And that's kind of what you're seeing these agents starting to get better and better at. The part that I think is so difficult about these legal workflows and similar to programming, is the boundaries between the tasks are super blurry. And so it's not easy to go to a law firm or go to a programmer and say, hey, the coding models

• 8:09Speaker 04

  or the legal models can do this, and humans can do this. Like, the boundaries are very blurred and the work is so complex, that a lot of the challenge we're working with law firms is how do you rethink your workflows and what humans and agents should be doing when you're working on these large projects?

• 8:24Ed Elson?

  When you talk about those boundaries, it's an interesting point, that the boundaries are blurry. Are you saying the boundary between what AI is best at versus what humans are best at, that it's hard to draw a distinction between those two things?

• 8:38Speaker 04

  I think it's both that, but also the distinction of even what do you delegate to humans?

• 8:43Ed Elson?

  Yeah.

• 8:43Speaker 04

  When you're doing a merger, you kind of have a pyramid, right? You have like a senior partner, some junior partners, senior associates, junior associates. There isn't like a concrete rule of when I'm doing a merger, this task goes to this associate. And there isn't even a concrete rule of what defines a task, right? Because it's all text based. And so it's a partner just saying, I need you to figure out X, Y, Z. And that could be something super simple, like go look up this one case and tell me who is the other party in it. Or it could be something super complex that is like, go write the first draft of the merger agreement. Right? But even that mapping, and this is kind of what you see with ChatGPT. Where when we started the company, people would be like, what does your product do?

• 9:27Speaker 04

  And it's kind of the same as asking people, what do you do with ChatGPT? It's like everything, but it's really hard to define why you do something one way. And this is exactly what makes the transformation so difficult. Cause to your point, given this kind of such an open natural language shape, how do you start defining the boundaries of this is what models are good at? Because it depends how you prompt it. It depends which model you're using. It depends on the agent harness. And so there's just this massive challenge of how do you organize all this work in this new way? Given the models can do some stuff, but they make mistakes in ways that aren't intuitive. And so it is just this huge change management problem and like up leveling problem.

• 10:09Speaker 04

  For not just legal, like all these industries, like you're seeing the same thing in programming right now.

• 10:13Ed Elson?

  Yeah, it does seem as though the great thing, what ChatGPT enabled us to do is ask questions that are actually blurrier and that that cannot be answered in binary, that cannot be answered that. I mean, it used to be that you had to spend a long time trying to phrase your, your question for Google search very, very specifically. And then what was kind of remarkable and liberating about large language models is that you could be a little bit more blurry and rough and it would be willing to go to those more ambiguous places and try to come up with more creative. Answers to more complicated questions. So on the one hand, I kind of think, well, that's exactly the strength of AI. So maybe this is exactly the place where AI should thrive.

• 10:59Ed Elson?

  But then, at the same time, you're also pointing out like, there are places where it still gets confused. I mean, large management work is still actually very complicated. And it's a lot more complicated in a large organization versus when you're just operating as a single individual, trying to figure out questions on your own time. I just want to point out for people who might be listening, because I mean, there are AI startups for everything now. And I'm sure there are probably hundreds of legal AI companies that are trying to eat your lunch at the moment, trying to compete. I would just note for people, I mean, from my understanding of the AI industry, Harvey is the number one AI company in law right now. You guys hit

• 11:42Ed Elson?

  a hundred and ninety million dollars in ARR in January. That's the most recent number we have. If you want to update it, go ahead. That was nearly double what it was five months earlier. So you guys are growing incredibly quickly. You are partnering with basically all of the biggest corporate law firms. You guys are kind of spearheading this transition. I guess the question then becomes, when you went to these law firms and you said, we can do what you guys do with computers. What did they say? Were they excited about that? Were they scared by that?

• 12:23Ed Elson?

  I mean, how did these big corporate law firms react when you, when you went up to them and made the pitch?

• 12:30Speaker 04

  So, I mean, the pitch is definitely not, we can do what you can do with computers. But I think what helped early on was we found kind of certain partners or innovation leaders that AI isn't new to law firms. They had been using things like TAR and other kind of AI technologies to do parts of legal work. I think this was just such a large step change. But early on, for example, our first client was A and O and David Wakeling there when we showed him kind of, we got early access to G P T four and we built a product around that and showed that to him. He just had this light, the same lightbulb moment we had where he's like, oh, this is going to change how we do work.

• 13:09Speaker 04

  And I think a lot of our pitch to law firms has been there will be parts of the work you do that these models will do, the same way now when you do discovery, you use TAR and use contract attorneys and you don't use associates. So that's going to happen. But there is also going to be a lot of work that these law firms do that these models aren't going to do, right? Like, I don't see a world in the next 10 years where you're doing a large merger or a large litigation and it's fully automated, right? Both for technology reasons, but also for regulatory insurance, all these reasons. And so a lot of the problem we want to work with law firms to help solve is what is the future of their business model going to look like, right?

• 13:51Speaker 04

  Because there are parts of this technology where you are selling expertise on an hourly basis, like there are parts of this that it will be complicated to figure out. There are new ways to collaborate with your clients. And so there's just going to be all these questions that this technology is going to raise for law firms, for all professional service providers, for most companies. And so I think a lot of the pitch is just, we want to be your partner and help you think through this entire transformation, not just the technology.

• 14:19Ed Elson?

  What you're essentially saying is, you know, it's what we can do is powerful, but not that powerful to the point where we're going to automate everything and basically eliminate all of these jobs. And you mentioned that there are technological constraints, regulatory constraints, and also insurance constraints, such that you couldn't just automate a big, complicated legal contract with your product and with AI. At the same time, though, someone like Dario Amade, who is leading the frontier of large language models over at Anthropic, is also saying that we could see roughly half of white collar jobs wiped out over the next several years. And so there is this tension where it's like.

• 15:04Ed Elson?

  You know, on the one hand, a lot of people in AI would say, might make the case that actually, no, you can do that. And I think a lot of people maybe listen to what you're saying now and they might be thinking, you're trying to tell us that it's not going to be that bad, so that maybe we could, people would be less afraid of your technology, less afraid of your product, maybe root for your product a little bit more. So I guess to press on that, what exactly are those constraints? Like, why, why couldn't you do this all with AI?

• 15:35Speaker 04

  I think the biggest is just the change management. So to be very clear, I think that technology is good enough now that it's like, maybe the timelines Dario is talking about, like, I roughly agree with those, but I think it's going to be close, it's going to be somewhere between what he's saying and like self-driving cars, where it's like self-driving cars are better than most humans at driving, yet they're 0% of the cars on the road, and they've been better than humans at driving for five years, right? It's like, it's clearly harder to roll out something like self-driving cars than this digital technology. But from the law firms we

• 16:15Speaker 04

  work with and the large enterprises we work with, I guess I'm still skeptical that like in a year or two, if you're working with like a large regulated bank, right? It's just like, I don't see a world where you just deploy these code models across the entire bank and say, like, the regulatory agencies just won't let you do this. And so I think there's going to be challenges like that. I think it's not clear to me how it plays out in terms of like, a lot of these arguments you could have made with computers and with the internet. And it's like, there was a huge class of jobs like, like brands and all these things. And it's like, oh, now you don't need this. And I think with like, a lot of the legal work that gets done,

• 16:58Speaker 04

  there is stuff that is just not digital, right? And so, like, from a capability perspective, I think I'm actually like, I did, I did AI research, I'm pretty aligned with like Dario and like, the way these things are going. And I think- To me, the biggest thing that I see in the legal industry is people don't think this is going to be as serious, or it's going to happen as fast as it is happening. Because we're seeing this happening in engineering right now, right? Like, these models are getting so good at programming that they are better than most human engineers, right? And to me, that's like, there's no clear research blocker that that's not going to continue. And it's going to keep speeding up, because I think that research path is pretty clear.

• 17:41Speaker 04

  And then a lot of what we're doing is, how do we take those models that are really good at programming and like, translate them into legal? And so I do think we will build systems that have the capability of like, automating large parts, most parts of transactions. My point is just, there's parts of that that are not technical problems, right? Like, if you think of a negotiation, there is a part of a negotiation that has nothing, it doesn't matter how smart you are, or how technical you are, right? It's like, it is a human-to-human thing. And it's like, when you meet the GCs of a lot of these large companies, it's like, when is the GC of a large private equity firm that is raising a $20 billion fund going to be comfortable with AI automating that entire fund, right?

• 18:25Speaker 04

  It's like, the downside's so big, it's just not worth it, right? It's like, okay, maybe you can do that 30% cheaper, but the cost of a mistake is like, that fund is structured incorrectly, and now I just lost $2 billion. It's like, I'll just have one of these law firms do it, and I'll have the partner review it, but hopefully they use AI for parts of it. And so I think there's just enough structural things where I think a lot of people think about legal as, can you review this contract? If you can do that, then we've reached automation. But like, what these large law firms are doing, and what these complicated regulatory industries are dealing with, is so much more complicated than just purely like, a capabilities intelligence problem. And so I think there will be things like that, that this is definitely going to happen

• 19:10Speaker 04

  on some timeline, I just don't think it's like, in the next two years.

• 19:16Speaker 05

  We'll be right back.

• 19:27Speaker 06

  If you're looking to hire, you need Indeed. With Indeed Sponsored Jobs, you can spend less time searching and more time actually interviewing candidates who check all your boxes. Less stress, less time, more results. And listeners of this show will get a $75 sponsored job credit to help your job get the premium status it deserves at indeed.com/podcast. Just go to indeed.com/podcast right now and support our show by saying you heard about Indeed on this podcast. Indeed.com/podcast. Terms and conditions apply. Need to hire? This is a job for Indeed Sponsored Jobs.

• 20:11Speaker 07

  This is advertiser content from Comcast Business. As cyber threats become more and more machine-based and driven by AI agents, those of us in the cyber defense space, we're having to invest at the same pace. Hi, I'm Chris McFarlane, Chief Development Officer at Comcast Business. So when we think about the threat of cybersecurity attacks to businesses, they're just a lot more automated. They have the ability to think, they can scale literally anywhere from 100 to thousands of time faster than a human can. The reality is that today, you need to protect every device, every endpoint, every workload that's running on a server, whether it's your own server or in the cloud. This is an area that Comcast Business excels at.

• 20:54Speaker 07

  We're enabling intelligence-driven defense with integrated visibility, AI-powered threat detection, and automated response capabilities that prioritize, correlate, and contain risks before they escalate. And so it doesn't matter whether you're a small business, or a medium-sized business, or a really large enterprise. We've got this incredible portfolio of cybersecurity solutions and services to really enhance your cybersecurity posture. To learn more about how Comcast Business can help protect your business today, visit ComcastBusiness.com/cybersecurity.

• 21:33Speaker 07

  This is advertiser content from Comcast Business. As cyber threats become more and more machine-based and driven by AI agents, those of us in the cyber defense space, we're having to invest at the same pace. Hi, I'm Chris McFarlane, Chief Development Officer at Comcast Business. So when we think about the threat of cybersecurity attacks to businesses, they're just a lot more automated. They have the ability to think. They can scale literally anywhere from 100 to thousands of time faster than a human can. The reality is that today, you need to protect every device, every endpoint, every workload that's running on a server, whether it's your own server or in the cloud. This is an area that Comcast Business excels at.

• 22:16Speaker 07

  We're enabling intelligence-driven defense with integrated visibility, AI-powered threat detection, and automated response capabilities that prioritize, correlate, and contain risks before they escalate. And so it doesn't matter whether you're a small business or a medium-sized business or a really large enterprise. We've got this incredible portfolio of cybersecurity solutions and services to really enhance your cybersecurity posture. To learn more about how Comcast Business can help protect your business today, visit ComcastBusiness.com/cybersecurity.

• 23:02Ed Elson?

  We're back with Gabe Pereira. I think the autonomous vehicle comparison analogy is a good one because it's true. It's like self-driving cars are technically better than humans at driving right now, or at least certain technologies will go with Waymo, for example. But at the same time, whenever there is an accident, it seems to be far more abrasive and scary and more concerning than when we see an accident, which happens at higher rates among the human population. And it sounds like maybe what you're describing is the same system where, I mean, you mentioned like a GC, one of the main...

• 23:47Ed Elson?

  Roles of that job, one of the main responsibilities is negotiation. Would you really let an AI do that for you? I would argue maybe you would. Maybe it's might be possible that actually the AI is better at negotiating. But it sounds like what you're saying is that if the AI were to make a mistake and if it were to eviscerate however many millions of dollars for the firm, that would be more difficult and more concerning and more of a problem than if a human being were to make a mistake. And therefore, we might be slower to roll this out than a lot of people think. Am I getting that right?

• 24:28Speaker 04

  Yeah, I think it's like the accountability is one piece. I think the other thing is like, so to me, the argument against why this goes faster than self-driving cars is like the obvious, you know, beyond regulatory for self-driving cars. It's just you need to build all these cars and it's like, it's hard to get all of them on the road. And so this will go faster because it's digital and you don't have that restriction.

• 24:49Ed Elson?

  Yeah.

• 24:50Speaker 04

  And I think to me, the biggest challenge is actually just how correlated all of the risk of adopting this technology all at once is. Like, if you think of what's so nice about self-driving cars is even for people not working in AI, it's quite intuitive what a self-driving car is, right? Like, you're just like, this drives, it didn't crash, I can kind of evaluate it. Accidents per mile makes a ton of sense. But now if you think of a bank, for example, adopting this technology, these agents, not just in legal, across the entire bank at the scale you would need to do the disruption you're talking about, you're taking on all of this correlated risk on this technology that makes mistakes in ways that isn't intuitive.

• 25:34Speaker 04

  And it's like company destroying if you take this risk incorrectly, right? And the same reason that self-driving cars freak people out, it's because they make, they cause accidents in ways that. You don't anticipate because you're like, as a human driver, I wouldn't make that mistake, even if statistically it's safer. And I think there's just going to be so many variations of that, where all of the second order effects and things like this. And if you think of legal, it's, this is like the underpinning of our entire society. And it's just like, does anyone understand the entire legal system and all the implications well enough? But if you just automated this whole thing in the next two years, we'd be like, that's gonna go well?

• 26:19Speaker 04

  It's like, it's just too complicated. And so to your point, will NDAs be fully automated in a year or two? Definitely. Will merger agreements for multi-billion dollar mergers? Like, I would be skeptical of that. And I think that's also the hard part of legal is there's just such a massive spectrum, right? And like, that's a lot of what we're helping law firms and enterprises figure out, where it's like, you wanna roll this out slowly, you wanna start using it on the low risk use cases so you can build this mental model, you wanna find pieces of the higher risk that you can take out. But it's not this super intuitive, just like, here's a bunch of legal agents, all your problems are solved. Like, it's the same way if you hired a thousand lawyers

• 27:02Speaker 04

  and you'd never run a law firm, you're like, I don't know how to manage them. And I think that's the big challenge everyone's facing.

• 27:08Ed Elson?

  It does seem a huge problem for AI that AI does make mistakes, and it still seems to make mistakes like relatively frequently. And on top of that, your point about an NDA versus a multi-billion dollar merger agreement. Humans make mistakes too, but at least you can tell a human, hey, you cannot fuck this up.

• 27:27Speaker 04

  Yeah.

• 27:28Ed Elson?

  You better get this right because there's billions of dollars on the line. And if you don't get it right, you're fired, or maybe you're gonna get into some legal trouble. I mean, you can really explain to a human the stakes of the problem and have some more assurance that they're not gonna make the mistake. It seems like you can't really do that with AI at the moment.

• 27:46Speaker 04

  That's one of the biggest values that I think people underappreciate of law firm partners. Right? It's like, at the end of the day, when you're doing one of these litigations or one of these mergers, there is like a single human being that has spent the past 20 years of their life doing all these similar transactions and being held accountable to that. And they're willing to bet their entire career that they're going to do that correctly. And it's like, that is a level of trust that I think we will get there with these systems, right? It's like, the stock market runs on like, an automated financial system that like, we now all trust. And so there's ways to get there. I just think there are a lot of problems we need to figure out to get there.

• 28:25Ed Elson?

  Just in terms of where you're at for the company right now, you you started the company a few years ago. You're now up to an eleven billion dollar valuation, nearly two hundred million dollars in annualized revenue. What are you seeing in terms of adoption on on on the front lines of these legal firms? I mean, how are they actually integrating it? And you mentioned earlier that, you know, for people within the legal industry, they don't think it's going to like massively disrupt the industry in the same way that people from the outside might. So talk a little bit about the adoption so far.

• 29:01Speaker 04

  To clarify the last point, I think most law firms and in-house legal teams now think this is going to change the industry a lot. I think my point is just with the, in the past couple of months, the jump with seam coding models, I think most of the world is still underestimating how much this is going to change everything. Like, I think everyone has somewhat gotten comfortable with what G P D four, G P D five, this like caliber of models means. I don't think most people have baked in what this next capability jump, which I think is equivalent to like G P D three to G P D four. And so in terms of adoption, most law firms we work with now, or all law firms are like, we need to adopt this.

• 29:43Speaker 04

  And I would say where most law firms are at is every one of my lawyers needs to be using this technology individually. There are a lot of law firms that are thinking about, here is how I need to change entire practice areas, like the workflows of a client matter that I'm doing with an entire team. And so starting to think about, okay, here's the parts of a merger that I'm just going to delegate to A I. And then I would say there is a decent number of firms that are then looking at the different ways they collaborate with their clients and price that work. And so. I would say the industry is definitely like moving in the right direction. People are starting to think about this and then their clients are also starting to look at this.

• 30:28Speaker 04

  So I think one thing that will be interesting is like most large enterprises have large internal legal teams and then also work with their law firms. And there's obviously like a blurred line between which work stays internal and stays external. And there will be this healthy tension of enterprises thinking about this is the work that I should do myself as these models get better versus this is a very specialized work that I want these outside law firms to do.

• 30:54Ed Elson?

  I just want to clarify your position on this because it sounded a little bit different from what you said earlier. I thought earlier you were saying within the law, the legal world, and we were talking in the context of job destruction. Within the legal world, it seems that lawyers, I mean, so from the outside, my, when I look at what's happening, I'm like, oh, this AI thing is going to completely transform the legal industry and it's going to destroy a lot of jobs. Destroy is a harsh word, but it's going to remove the need for a lot of jobs. And it sounded like you were saying, well, actually, within the legal industry, when we look at what's happened so far, people in law are less worried about that than you might be. But I'm also hearing at the same time that you're, you're pointing out the capabilities of these new models.

• 31:37Ed Elson?

  We've just seen what happened with, with Anthropic's new model, Claude Mythos, which is just, I mean, eviscerating stocks across the board. And you're saying, yeah, people aren't seeing how much, how transformative this is really going to be. So I guess my question is like, where do you, where do you really stand on this? Do you think that we are overestimating the impact or underestimating the impact specifically when it comes to law?

• 32:01Speaker 04

  If we go to the extreme, do I think 50% of these jobs will go away in the next two years? Like, I would say that's too extreme. And then I would say. Where are law firms right now? I would say most of them are underestimating to your point of like how good this technology is going to get. And then I think there is, from a purely capabilities perspective, I think these models are like exactly what you're talking about of like Mythos. These coding models are incredibly powerful. I think there is a lag of the effort to productionize that in a way that these law firms can use. And so I think part of the capability gap you're seeing right now is like,

• 32:42Speaker 04

  the reason programming is happening so fast is there's basically no implementation cost. Right? Like anytime a new model comes out, I can just go in terminal and I can be like, oh, Codex 5.4, extra high, fast. Let me just use that and swap models and I can use it on my entire code base. And it's like very easy to like absorb all the new capabilities. The lag we're seeing with law firms is you can't use desktop products. Right? Like if I'm working at a law firm, I'm working on an internal investigation for Goldman Sachs. I'm not allowed to download that data onto my desktop and use a code model on it. Right? And so even though there is, to your point, this massive capability jump,

• 33:24Speaker 04

  I think there is still a lag to deploy this technology into a law firm and enterprise. Yes. A lot of what we're building now is all of the like security and things that you need to be able to use all of these capabilities in like a controlled way. Or just like a simple example, there's all these examples of use the code model and you're like, hey, can you change this song on Spotify? And it's like, oh, I don't have an API. I just like went on your desktop and wrote this Apple script to like get around these restrictions. Right? But if you're doing that for like a sensitive merger that's not public and it's like whoops, emailed this to the wrong person. I think there's all these things that you need to do to constrain it. And so I think that will slow this down a bit.

• 34:06Speaker 04

  But I, to your point, like from a purely capabilities perspective, like yeah, these models are like senior assistants. Like they're just incredible. The shape of them is just still weird. The same way like the coding models aren't quite there where it's like. We can't have them at the scale we're at. They don't quite architect our product correctly, right? If I'm just like, go build this new product at the scale we need and make all the right architectural decisions. They don't quite do that. But if you take like a principal engineer in these models, the stuff you can do is insane. Like there is senior partners that I talk to that are working on very large mergers that are just like, I'm able to do most of this merger with me and the model and doing the work.

• 34:49Speaker 04

  But I think the other thing that is going to make this go slower is these models are very powerful when you know how to use them. And so what you see in programming is because the code models are just so perfectly aligned with the way you do engineering. Like most engineers are grocking how to get the like value out of these systems very quickly. There are very few lawyers that are understanding these models the way like engineers, because it's just not as intuitive. And so I think like, to answer your question, because I think it's complicated, like in two years, do 50% of these jobs just all go away? I don't think that happens for the reason I said. From a capabilities perspective, if we could perfectly diffuse this into the industry,

• 35:33Speaker 04

  can it do 50% of what people are doing today? My guess is probably yes. But I think that diffusion, for all the reasons I talked about, I think happens a bit slower than maybe like Dario is predicting in terms of just like, this happens next year for regulatory security, all of these reasons. And then the last point is, I do think law firms in general think this is going to happen slower than I'm saying.

• 36:01Ed Elson?

  We'll be right back.

• 36:11Speaker 08

  This is advertiser content from IBM. Cold take. My laptop works fine at room temperature and most things on earth do. If quantum computers can only function and behave quantumly at insanely cold temperatures, how will they ever become useful to most people?

• 36:29Speaker 08

  My name is Olivia Lanes. I am the Global Lead for Content and Education at IBM Quantum. I want to emphasize that actually cooling down a quantum computer is not really as hard as I think people think it is. It really is just the push of a button that can take your entire infrastructure down to 10 milli-kilovan. That is not the challenge, scientifically speaking. I also want to emphasize that in the future, we don't envision everyone having their own quantum laptop. Quantum computers are going to live in a data center. Data centers already have their own cooling infrastructure. It's not going to be something that you have to learn how to install and work on a personal

• 37:11Speaker 08

  level, but that doesn't mean that the discoveries made with the technology won't affect every person. Quantum computing has the potential to touch almost every industry that you can probably think of. If you would like to learn more, you can do so at ibm.com/quantum.

• 37:29Speaker 05

  Is your house in absolute flames right now? Then you need the Fire Advisory Squad! At Fire Advisory Squad, our trained firefighters assess your fire and provide hundreds of real-time suggestions on how you can extinguish it.

• 37:45Speaker 00

  Just add water.

• 37:46Speaker 05

  No fire is too big or too small for the Fire Advisory Squad, because we're just sliding up and down the fire pole for fun over here. You're the one hosing things down. Careful, fire is hot. Booyah!

• 38:00Speaker 09

  Fortunately, Fire Advisory Squad isn't real. That would be ridiculous. Kind of like AI that suggests what your employees should do but can't help them do it. That's like a firefighter that only tells you how to put out your own fire. ServiceNow doesn't stop at suggestions. It puts AI to work, connecting your AI, workflows, data, systems, and governance to take action and get the job done. That's why the world works with ServiceNow.

• 38:30Speaker 10

  There are sales teams out there who never seem to stop moving. Always a step ahead, always working with a level of precision that doesn't seem normal. It's no secret they're on Attio, the Agentic CRM. Attio runs the work behind every win. It's built for humans and agents with the guardrails of human software and the scale of Agentic infrastructure. This is the frontier of go to market. Teams like Parallel, TurboPuffer and Wordsmith are already using Attio to set the pace. Now it's your turn. Try Attio for free by going to attio.com/fox. That's A-T-T-I-O dot com slash fox.

• 39:20Ed Elson?

  We're back with Gabe Pereira. So I very much agree with the points you're making here. And this is something that I found kind of interesting. And I think the market is getting it wrong on the SaaS apocalypse that we're seeing because we see these incredible models come out. They're incredibly smart. They're incredibly capable and Salesforce gets battered and CrowdStrike gets battered and CloudFlare and ServiceNow and all of these, all of these software enterprise CRM management companies just, just getting battered right now because the idea is like, okay, Anthropic can do, can create the products that they do. But this, the thing that they seem to be forgetting and that seems to be a very, very

• 40:02Ed Elson?

  big piece of the puzzle is you, there are other things that are important when you're managing a business that aside from capabilities. One of them would be trust. One of them would be security, cybersecurity. One of them would be privacy. Another would be relationships. I mean, those are the kinds of things that seem to matter a lot across basically all forms of white collar work where yes, there might be some vibe coded alternative that can do the job of the associate pretty quickly. But if you can't trust that it's going to protect the data, or that it's going to do something kind of crazy, or that you haven't even gotten the regulatory approvals,

• 40:47Ed Elson?

  or you haven't even got the approvals from your data partner, or whatever it might be, then you simply cannot use that product. And it seems as though that, I mean, from a product perspective, maybe you would argue, yeah, that's not Salesforce's strength right now. But in terms of privacy, cybersecurity, the relationship with the customer and the client, that seems to be the real strength. And it sounds like what you're saying is like, yes, these models could do a lot of the work in the world of legal, but there are all these guarantees and all these other parts of the business relationship that they can't do yet. And so they're just not, they can't use them at the moment. Or at least, they can't use them in the extent that a lot of people seem to think.

• 41:32Ed Elson?

  I guess, would you agree with that characterization?

• 41:34Speaker 04

  Yeah, I think that's right. Where it's like, I think what's going to happen is the models will do much of the, like, if you're an associate at a law firm, like, you are, for the most part, not interacting with the client, right? You are doing all the work that the partner delegates. And I do think the models will increasingly do more of that. And you will need to build these hybrid law firms that are, here is a ton of agents and probably less associates that do all the things that the models can't do. And there is a huge amount of legal work that is that. And so I do think, like, there will be a lot of legal work that these models can do. And then I think the point you made, that I think it does feel like people miss of like, what is the value of these enterprise SaaS companies? Is exactly what you said,

• 42:18Speaker 04

  where it's like, there is a huge difference between vibe coding a product and building a product and an organization that another company is willing to bet their company on you.

• 42:28Ed Elson?

  Yeah, it's what I put.

• 42:29Speaker 04

  Like, if Salesforce goes down, your, like, sales org can't function, right? And like, this is the same in legal with the private equity example I gave. Like, if you want this to do a fund formation, I need to wait 10 years until that fund pays out to know that you structured this correctly. And it's like, These systems are getting so complex that you can't evaluate them, right? Like, if you think of, how do I evaluate that a partner is a good partner? There's no test I can give that partner. It's just that person has done mergers for the past 20 years and for the most part, those mergers have gone well. And it's the same with software, right?

• 43:11Speaker 04

  Like, I think anyone who's built software, it's like you can test stuff, but at the end of the day, it's like, did your system run in production for 10 years and not go down? And there's no test for that. And all of these things are ways that you build trust of, like, you can't shortcut the trust. That, to me, is going to be one of the biggest things that slows this deployment of this technology. Because especially if you say, I'm gonna have one company or one model build all of this, you're just taking all this correlated risk on that single system. Where it's like, if one thing is wrong with that model and it wrote all of your code and all of your infrastructure, and you don't understand it, like, your company is over.

• 43:53Speaker 04

  And so I just don't think banks, private, like, you just can't take this correlated risk. And that's the value of something like Salesforce CloudFlare. It's like, you have now spread out this risk to a bunch of these different parties that are accountable for very important, but like, separate, uncorrelated parts of your business. And it's like, that, to me, is, I think, when sometimes people talk about, oh, we're just gonna have one model that solves everything. It's just, the world's too complicated, I think, for things to play out that way.

• 44:24Ed Elson?

  I think this is a pretty good segue into the next question. Your company relies on the foundation models from companies like Anthropic and OpenAI and XAI. I mean, I guess it's like, there's literally just a handful of companies that are actually building the foundation models. And that is the case for basically every AI startup. I mean, I know founders who are building AI for finance and who are building AI for all these other industries, and they all rely on the foundation labs. And the question is always to those guys and to companies like you, to a founder like you, what is stopping? Anthropic from killing you. What's stopping OpenAI from killing you? If you are literally dependent on their models

• 45:09Ed Elson?

  and they appear to have an ability to build the models themselves and to implement them into various industries. I mean, OpenAI can build codecs and they can just insert that into the engineering, the software engineering space. What's stopping them from doing that with legal?

• 45:25Speaker 04

  I mean, I think all of these companies are going to do that. And I think to me, the tension is, I think there's actually a lot of companies building these models, right? It's the foundation model companies you mentioned, plus all the cloud providers, where they either have their own or you can get the foundation models through the cloud provider. And so I think you've kind of distributed that risk. And then to me, one way I would think about it is, like, why didn't Google build a data room product? Right? Like, data rooms in the previous generation of software was like a multi-billion dollar industry. It's basically just Google Drive, right? But no one uses Google Drive to do a transaction. There's like an entire industry of companies that build data rooms, right?

• 46:08Speaker 04

  And if you think of the biggest challenge for these companies is, how often does Dario or Sam think about the legal industry in terms of building a product? Never, right? It's like they're thinking about, how do I compete with hardware? How do I get the funding I need to build data centers? How am I going to compete with the cloud? It's just like, no part of that company is thinking about that, right? It's like, and then when you look at Microsoft or these large organizations, it's like, they have a small GTM team that sells some of their products to law firms. But the thing that I think people don't understand is like, the problem you need to solve for law firms is not just Can this model do legal work?

• 46:53Speaker 04

  Like, this entire industry is about to go through a transformation where the way you structure your firm needs to change. The way that you build clients needs to change. The way you train associates needs to change, right? Who's gonna help these law firms and their clients go through that transition, right? And that, to me, is like the problem we're solving, where it's not just who can build the best models. It's what is the platform and all the change management? And if you think of companies like Salesforce, like, that's the value they provide. Where, when you need to build a sales organization, like Salesforce just has so much of that learning from helping every company do this. And I think what's gonna happen will be similar to what happened with cloud, right?

• 47:37Speaker 04

  The models are going to be, become a core part of like societal infrastructure. Like, that's very clear now. And these model companies are going to build massive businesses selling these to companies. And then there will be parts of products they can build, right? It's like cowork, ChatGPT, it's like, these are incredible products that are very horizontal. But I just think these industries and the world is so complex that is there very large businesses you can also build on top of these platforms? Like, I think for sure, right? Like, this is what you've seen with every other platform shift, with the internet, with computers, with mobile phones, with cloud. Like, the reason these things become such large companies and such powerful technologies

• 48:19Speaker 04

  is because they are platforms that enable companies like us to build massive businesses on top of them. Like, that's almost what necessitates them being able to capture so much revenue. And then I guess the last piece I would think of is like, right now, what's stopping anyone from starting a company, right? Like, anyone can go hire a bunch of people and coordinate them, but it's actually, turns out to be quite hard to do this. And I think that's, I think that's what you're gonna see with agents, right? Like, very soon, every person is gonna have the ability to hire infinite employees. And it's like, this is going to hugely democratize people's ability to build companies. And there will be really valuable small companies that are super specialized.

• 49:02Speaker 04

  And then there'll be people that figure out how to do this at scale. But I just think this opens the pie so much that it's like, there's just no world where just one provider does all of this, right? Because if the argument is like, oh, they do this for legal, then presumably they do it for every other industry in the world. And it's like, I just don't think that's how this plays out.

• 49:22Ed Elson?

  It sounds like one of the things that we're learning here is that the one of the biggest shortcomings of AI right now in terms of actually implementing it into the framework of an enterprise is trust and security. And that puts you in an interesting position because your company is literally three years old. And so the idea that you're going to come to these big law firms and say, don't worry, you can you can count on us. I I worked at Meta, I worked at DeepMind, I'm a young guy, and this is my roommate, and you can trust us with your data. That seems to be a pretty bold, a bold statement. How have you navigated as a founder and as a pretty early first time founder,

• 50:01Ed Elson?

  like, dealing with trying to get people to trust you? Like, how do you actually do that as an entrepreneur?

• 50:09Speaker 04

  The thing definitely that helps now is, I think we were telling law firms about this before any of this happened. Like, we started the company before ChatGPT. A lot of the things that we have told law firms, investors, enterprise, like, a lot of these things have come true. And I think that is a big way you build trust over time, right? The things that you say are going to happen or that you say you're going to do, like, you do those. So I think that's one piece. I think the second is the team you're able to build. And so I think we've just built an incredible team. And it's like, Winston and I are relatively young, but if you look at our C-suite, these are people that our CTO has, Siva has managed a thousand person engineering org,

• 50:53Speaker 04

  our CLO has taken a company public. And so we've built a team, not just the leadership, the entire team, where I think you can trust them. But I think to your point, it's we haven't fully won this trust. And so there is a ton of work of, we need to build the best product, we need to build the best team, we need to keep scaling, we're partnering with kind of all the other providers. And so we work with the existing legal technology companies and enterprise companies. And I think the more that you can just work with the entire industry, like, that is how we build trust over time. But I think to my earlier point, like, you can only do it so quickly. And so for us, this is, you know, a ten, twenty year company, and we're just going to keep doing it.

• 51:36Ed Elson?

  How much does branding play a role in all of that? Because something I think about with this is, What you really want, especially for these very storied white collar institutions, is you want to present as institutional. But if you are a startup, you are by definition not institutional. And that seems to be the real problem for a lot of companies that are trying to break into these very institutional industries, is that this isn't a place for startup. This is law. This isn't a place for technology. This isn't a place for young guys who are founders. At which point, it seems like part of the job is to be like, no, no, we, we have an institutional credibility. And I would imagine that a lot of that falls on the, on the responsibility of the brand side. Is that true? And how have you thought about branding?

• 52:20Speaker 04

  I think one thing Winston and I thought a lot about branding is by all the things that you don't do. And so I think we got kind of a bunch of flack early on because we like didn't do a bunch of marketing, didn't talk about what we're doing. And I think one thing that always inspired us was when you look at the websites of these top law firms, like they never do any marketing, right? Like if you look at a top transactional law firm like Wachtel and you go on their website, all you see is the caliber of their team and the size of the transactions that they've done, right? And they let the work speak for themselves. And I think when we think about our brand, like that is very much the brand we want to build, where I think one thing we take great pride in is all of the law firms and

• 53:04Speaker 04

  the enterprises that we associate with. And that to me is one of the biggest ways that we are able to build trust and show this, where we have gained the trust of, you know, these Fortune 500 companies, these top law firms, and we've worked with them in a way where they speak highly of us. And like, to me, that's the ultimate way to, to build trust. I think a lot of times when people think of brand, they think of kind of like marketing and design. And I think those things matter, but when you think of an institution, it comes more from like the things that I'm talking about and less of like how you design the product.

• 53:36Ed Elson?

  As someone who has built an extremely successful company, $11 billion valuation for the entrepreneurs and the first time founders listening to this podcast, what advice would you give them?

• 53:48Speaker 04

  I think the biggest right now is just use the models. Like I, like if I was right now to, you know, start a company again, I would just be using the coding models 24/7 because I think. To me, I would say the big opportunity coming is I think the company we started seems obvious now, but at the time wasn't obvious. And I think the companies that will be successful starting now are the ones that don't seem obvious now. And it's like going and doing legal or any of these verticals, I think seems somewhat obvious now. But the thing that is not obvious now is I think when people, when we invented the internet, no one anticipated Uber, TikTok, DoorDash, these companies. To me, the really interesting

• 54:30Speaker 04

  startup question is like, what is the shape of those companies on top of generative AI? And so I would say like, use the models and, you know, figure out what those are. That that to me feels like the big, the big interesting question.

• 54:43Ed Elson?

  Gabe Pereyra is the co-founder and president of Harvey Gabe. Appreciate your time. Thank you.

• 54:48Speaker 04

  Thanks so much for having me.

• 54:53Ed Elson?

  This episode was produced by Alison Weiss and engineered by Benjamin Spencer. Our research associates are Dan Shalane and Kristen O'Donoghue, and our senior producer is Claire Miller. Thank you for listening to the Prof G Markets Founder Series. We'll see you next month with another founder story.

• 55:21Speaker 11

  Running a business shouldn't feel like surviving a software group project. One app for accounting, another for inventory, another for sales, and somehow, none of them talk to each other. That's where Odoo comes in. An all-in-one business management software that brings every part of your business together. From sales and accounting to inventory and marketing, all in one powerful platform. No messy integrations, no bouncing between tabs, and best of all, no spreadsheets. Stop managing software and start managing your business with one unified system. Try for free today at Odoo.com/vox. That's O-D-O-O dot com slash vox.

• 55:59Speaker 12

  Trying to locate news sources that reliably separate fact from fiction can seem like looking for a needle in a haystack. That's why The Guardian is launching Stateside with Kai Wright and Carter Sherman, a conversation with experts who slow down the news and wrestle with the questions we all have about what's actually happening in the world. Three times a week, hosts Kai and Carter utilize all the reporting resources The Guardian has to discuss the news, international affairs, climate, culture, sports, lifestyle, fashion, and wellness. And The Guardian is not billionaire-owned, meaning they're free to report the whole picture without interference. Go to theguardian.com/stateside to learn more. And listen wherever

• 56:41Speaker 12

  you get your podcasts, or watch on YouTube. That's theguardian.com/stateside.

• 56:50Speaker 13

  You're listening to DeepGram Flux TTS. Different voices, same model, all ready to speak.

• 56:56Speaker 14

  Flux TTS is a streaming text-to-speech model built for voice agents.

• 57:00Speaker 13

  Flux reads the room. It holds context across the conversation, turn after turn.

• 57:04Speaker 14

  With consistent tone, interruption handling... And plenty of personality.

• 57:09Speaker 13

  So your agents keep it flowing and customers can just keep talking.

• 57:12Speaker 14

  Try DeepGram Flux TTS free now until September 12. Visit deepgram.com/keeptalking. Terms apply.