Gabe Pereyra, Harvey, and Ilya Fushman, Kleiner Perkins: How law firms went all-in on AI.

0:00Hello and welcome back to the Scaling Europe show. I'm Seb Johnson. Thank you for following along. Today we've got a great conversation from our time at Slush and the conversation is supported by Databricks, whose unified platform power source on everyone's FY2026 target, becoming a truly AI-driven company, and Harmonic. Harmonic is the startup discovery engine used by leading VCs across the world to find their fun returners. Check them both out. Thank you. Hello and welcome to the Scaling Europe show. I'm Seb Johnson. We are here at Slush in Helsinki and we have got two amazing guests in front of us here.

0:32We've got Gabe and Ilya. Gabe, co-founder, president of Harvey. Doesn't really need an introduction. You've had an absolutely phenomenal year. It's insane to believe that you're only three years old. One of the leading AI startups, if not the leading AI startup in legal tech, and Ilya, partner at Kleiner Perkins, who, of course, invested in Harvey. Now, before we talk about the investment in Harvey and tech more generally, how did you two get to know each other? How did that investment come about? Yeah, we actually met, truly met at the Series B, which was the round that we wound up leading where I wound up investing.

1:06And in fact, I think for the first time we really met when you guys came in to not pitch pitch us at the office. It was a not pitch pitch. There was no deck. It was mostly, I'd say, vibes. But the vibes were good. But we, of course, knew about Harvey for a very long time. We'd known about Gabe and Winston, about the growth of the company. And they'd been on our radar for a very long time. In fact, we were like a day late for the Series A, I think. So that's always been something that we've talked about. Yeah, and I was going to say right after the not pitch pitch that we had, I think maybe three hours later you called us and said we're in.

1:43And I think that was kind of when we knew we would work super well together because we, Winston and I make decisions super quickly and like wanted investors that did the same. And how important is like the vibe that you have with an investor before kind of letting them in? I think super important and more than maybe vibe, just I think Winston and I, all of the early hires, leadership hires, very quickly, we've been able to just within the first early conversations, you just know if you're going to work well with something. And I think we had the same with Ilya and our other investors. So I think that's super important.

2:18You can kind of just tell. Yeah, yeah. And what about Harvey kind of, you know, made you so quick and eager to get involved? Well, Harvey as a company had been on our radar for a very long time, because if you think about applications of AI to work more broadly, which is something that we think about quite a bit and we thought about as kind of our entry point to investing, just given the history of having invested in companies like Slack and Rippling and Intercom, the idea that you would have a co-pilot or some kind of experience, this is in 2023, for every professional in their life made a ton of sense.

2:52and if you sort of looked at the list of those types of things it would be software engineering it would be doctors and would be lawyers and professional services so the concept of building of having something like Harvey made a ton of sense and specifically the reason it makes sense is that the work is incredibly difficult it's incredibly nuanced it's incredibly subjective it's got a really interesting market structure to it and so the the Harvey approach and the way that they really focused on the market really resonated with us on those dimensions. Super interesting. And, you know, the kind of Perkins seems to have a really strong thesis about vertical AI.

3:27Like you mentioned, you've made investments into Harvey, but also AI tools for doctors, for engineers. Are there any other professions that you're looking at right now that you think are very ripe for that type of model, be it accounting or teaching or other kind of similar professions? I think the quick answer is really it's going to be every profession and every one of us is going to have, you know, if I think about my kids, they'll have co-pilots, right? They'll have a friend, they'll have a tutor, they'll have an educator. So I think it's going to be pervasive. The real question then becomes, how do you build something that is significant and standalone?

3:59And how do you really get it to work in a way that creates the value? And I think that's been actually really a function of development of models and performance of models and development of product around that. you know it's but every truly every role like every you know professional services legal services doctors lawyers nurses services companies customer support agents sales folks like and we have we have companies and kind of doing things for all those categories but it's really everyone yeah nothing's off the table i guess nothing's off the table and as model performance improves as the cost of infrastructure go down you can reach broader and broader audiences with these products and can I can I ask Gabe we talk about models improving costs coming down how does that change the core product at Harvey how does that change the product roadmap how do you think about it when models get released I think this has been something from when we started the company we always had the bet that the models are going to get much better than they are even today they're going to get much cheaper and so really thinking about what is the value that you're providing.

5:04And when we think about the law firms we sell to and the enterprises we sell to, at the end of the day, if you make every associate a bit more efficient, this doesn't actually improve necessarily the business of the law firm. And so we really think about how do we make these law firms more profitable, not just every individual employee a bit smarter. And same with these enterprises with massive legal functions. How do we help them get better, faster, cheaper legal services. And I think at scale for these enterprise companies, it really is the agents, the models are part of it, but it is all of the product, the governance, the process management, how you work with the humans in those organizations.

5:44And so that's a lot of the problems we're starting to solve. And how did you work with those very initial lawyers? I mean, you first started Harvey, you're building the technology for what is a quite traditional industry. How did you sit alongside, work for and build for lawyers? Yeah, I think a lot of the intuition came from, so my co-founder Winston was a lawyer and so it was just me and him sitting down using these models. He had been an associate at Melvany, which is one of the top litigation firms and just going through the work that he does and figuring out here's what the models can do, here's what they can't do, here's all the product you need to build around the models and then as we scaled up and started working with large law firms, it was, here's what teams of lawyers need to complete these complex projects.

6:28Here's what admins need to manage their entire workforce and these kind of agents they're deploying. So very close to the user experience and yeah, that classic product iteration, St. Kletcher customers. I read that Harvey's first customer was actually a UK customer. Can you touch a bit on kind of what happened there? How did that come about? Yeah, so one of our early investors, Sarah Guo, introduced us to an ex-A &O partner who was doing an MBA at Stanford. And when we pitched him the product, he said, you need to meet David Wakeling, who was a senior partner at A &O, but also ran kind of the technology side of the law firm.

7:08And when we showed him the initial product, he was like, this is what I've been looking for in legal tech for the past 10, 20 years. and full credit to him he recognized what this was going to become and we went very quickly from an initial pilot of 30 people to a firm-wide rollout of 3 ,000 lawyers really early on and I think that was the moment where we were like okay there's something very big here. That's amazing that's the hearing somebody say this is what I've been waiting for for 10 years is like the dream product feedback and how are you thinking more broadly about Europe European expansion signing more and more UK European clients?

7:46Yeah and so our second customer after them was PWC Europe and so I think from the start we have been European focused and then international so we're in 56 countries now and so I think a lot of the focus is how do you localize the product so how do we get case law in every country every country's legal system is different the way they practice is different and so really thinking about beyond just different languages how we make the product work in all these different countries and then i think a lot of the value for these large enterprises that operate in all these different countries you need to work with a network of law firms a lot of these law firms have kind of sister firms or partner firms in other countries and so starting to think about how we can tackle all these problems globally yeah that's a really interesting like go-to-market motion if you can sign one office in london who's got offices all over the world it's a really nice entry point i also want to talk about your background and your experience because you you were at DeepMind right DeepMind is you know one of our UK's you know almost like proudest achievements the AI before AI was really big how formative was your experience at DeepMind in the approach that you took to building Harvey?

8:52I think the probably the biggest thing I took away from it is even when we got early access to GPT-4 we were already thinking about agents. So I had done a lot of reinforcement learning stuff at DeepMind. And when you saw GBD4, it was very clear, you're going to connect these to tools, you're going to be able to give it these very high level tasks, like go do a diligence, and the models will be able to just figure out how to do that. And so we've been thinking a lot about how do we build all of the product and infrastructure around these models, so we can do this more complicated RL stuff. And then I think also just having spent that time in Europe, the first startup that I actually tried to start was here in Europe.

9:35Oh, amazing. And I think a lot of the network from that as well is super helpful. Yeah, amazing. And Anil, how are you and Kleiner Perkins thinking about Europe? I mean, we touched on the very, almost like, yeah, AI verticalized, the application layer of AI that Kleiner Perkins is going big on. That's an area that we're really over-indexing on in Europe, you know, but we're investing a lot of money in the application in Europe. How are you thinking about the European tech ecosystem? Well, first and foremost, I think Europe has incredible talent, right? I think, and Europe has a history of incredible companies that have been built here.

10:05And so we're quite excited about Europe as a birthplace of great companies. And we've got, you know, Harvey's expanding in Europe. We have Synthesia, which has also got European presence. But we have actually quite a few companies where either the founders came from Europe or they're kind of expanding into Europe. I used to live in Europe too when I was growing up. So it's always fun to be back. But I think, so our view is there's incredible talent, and our goal is to find that talent at the right time and help that talent grow. And typically for us, that means having the company have some presence in North America, in the U.S., where we can be hands-on helpful.

10:44But, you know, I think you've got incredible application-level talent, but you also have incredible research labs that will eventually produce more applied AI, more infrastructure AI type of opportunities. So yeah, quite bullish and excited to be here. And can I ask, very open, honest, no polite answers. You know, the media company that I've built is all about taking European tech to the next level. What are we doing in Europe that is wrong or bad? Or that people in America are looking at and thinking, what are you doing? I don't think there's anything. So first, I don't think there's anything.

11:19There are some things that make it harder to build companies here. Absolutely. Right, and I have investments in companies that are GmbH structures, and there's governance, and there's all sorts of things. Equity is not thought of the same here as it is in the States, so kind of incentives. But fundamentally, I think the biggest thing is having a crop of incredible entrepreneurs who kind of recycle back into the system. So I think in some ways it's a little bit of a waiting game, or it's a game of founders coming over to the U.S. where when you need to hire your star-studded CTO, VP of engineering, chief people officer, chief operating officer, chief revenue officer, just the pool of those people is right there.

12:00And you have other founders that you can kind of bounce. That community, I think, is a little more dense right now in the States. And so I think you kind of want to, over time, recreate that kind of community in Europe. And I think that's really what you need because you just need the right people to build these incredible companies. Yeah, absolutely. We definitely need more people in tech generally and more people who are willing to spend their time taking the path that you took and going deep on the research level and then spinning it out into a business. How do you think about global talent?

12:30One of the things that I speak to a lot about with European founders is how a lot of the product engineering talent is great in Europe and super cheap compared to the US. And San Francisco has got amazing talent, but it's super expensive. you know a global company in you know 50 plus countries how do you think about talent where to hire people where to base them yeah i think we're still at the stage where for us it's just how do we get the best people and i think you find them everywhere in the world so we just recently hired one of our engineering leaders in toronto who had worked with our cto siva at twitter and he's phenomenal and it was just worth having an office there to get him on board and then we are starting to hire folks in London and so I think there's a bunch of kind of like Gilly said a bunch of good research talent folks from DeepMind and just more broadly I think there's starting to be a lot more technical talent and then we're also opening an office in India and so I think we're trying to find more and more talent out of SF especially as we scale but I think you can find it everywhere.

13:37Yeah, amazing. And it makes total sense. I want to finish with one question each. And I'm going to ask Gabe a slightly worse question in the sense, and you're going to have like the opposite question. What has been the hardest or lowest point during your entrepreneurial career where you felt like you're up against a brick wall and you don't know how you're going to get through it?

14:00I think it's definitely, to me, the challenge has just been the rate at which we're scaling. And so I think for me, the hardest part was my background is AI research. And so I spent 10 years kind of in a lab by myself without deadlines, without meetings, without people just thinking about AI and I think the transition from that to the past three years of scaling to we're now over 400 people over a thousand customers just that shift I'm like pretty introverted and so I think just learning to deal with that learning to operate in this new way I think that's been really hard but I think having folks like Ilya my co-founder kind of the team we've built has helped but there's been times where I think that's yeah no that's a really interesting insight yeah I mean there's not that journey from sort of researcher, super technical to star-studded co-founder is, yeah, it's a really interesting journey.

14:57And I know you've got an amazing track record, you've backed companies like Rippling, Intercom, and of course Harvey. What has been the defining deal or moment of your career that you look back on with the most joy or most pride? Besides Harvey. I think it's obviously Harvey. You can't say Harvey. You know, I think I would say the best might yet be still to come. I think we live in such a I mean, if you think about the moment we live in today, it's bigger than the Internet. It's more incredible. It's stronger technology. It's going to be more leveraged across pretty much every facet of what we as humans do, obviously, from the day to day to through the research and discovery of new things.

15:39so I think honestly the most exciting thing is like the stuff that will be built in the next five to ten years I think is going to be more mind-blowing than anything that's been built today but of course I expect for Harvey to be one of those next in line you know Magnificent Seven type of companies because it's such an opportunity but I think that's what keeps me kind of going and excited. That's an amazing answer you know I've asked a lot of you know I'm always looking to see the differences between the European ecosystem and the American ecosystem to see how can we replicate the success that so many American tech companies have.

16:12And I think that mindset that you've kind of portrayed there of looking forward and being super optimistic and ambitious kind of really captures the spirit that you have in the US. And lots of investors that I talk to are more backwards looking and they talk about the deals that have happened. But I think that's an amazing answer that really captures a lot of the mindset difference and the ambition difference that we have. Well, thank you so much for joining me. It's been an absolutely pleasure to chat to you both and best of luck with your speaking slot. Thank you. Thanks so much for having us.