This week, we talk with Gabe Pereyra, President and co-founder at Harvey, about his path from DeepMind and Google Brain to launching Harvey with Winston Weinberg; how a roommate’s real-world legal workflows met early GPT-4 access and OpenAI backing; why legal emerged as the right domain for large models; and how personal ties to the profession plus a desire to tackle big societal problems shaped a mission to apply advanced AI where language and law intersect.
Gabe’s core thesis lands hard, “the models are the product.” Rather than narrow tools for single tasks, Harvey opted for a broad assistant approach. Lawyers live in text and email, so dialog becomes the control surface, an “AI associate” supporting partners and teams. Early demos showed useful output across many tasks, which reinforced a generalist design, then productized connections into Outlook and Word, plus a no-code Workflow Builder.
Go-to-market strategy flipped the usual script. Instead of starting small, Harvey partnered early with Allen & Overy and leaders like David Wakeling. Large firms supplied layered review, which reduced risk from model errors and increased learning velocity. From there the build list grew, security and data privacy, dedicated capacity, links to firm systems, case law, DMS, data rooms, and eDiscovery. A matter workspace sits at the center. Adoption rises with surface area, with daily activity approaching seventy percent where four or more product surfaces see regular use. ROI work now includes analysis of write-offs and specialized workflows co-built with firms and clients, for example Orrick, A&O, and PwC.
Talent, training, and experience value come next. Firms worry about job paths, and Gabe does not duck that concern. Models handle complex work, which raises anxiety, yet also shortens learning curves. Harvey collaborates on curricula using past deals, plus partnerships with law schools. Return on experience shows up in recruiting, PwC reports stronger appeal among early-career talent, and quality-of-life gains matter. On litigation use cases, chronology builders require firm expertise and guardrails, with evaluation methods that mirror how senior associates review junior output. Frequent use builds a mental model for where errors tend to appear.
Partnerships round out the strategy. Research content from LexisNexis and Wolters Kluwer, work product in iManage and NetDocuments, CLM workflows via Ironclad, with plans for data rooms, eDiscovery, and billing. Vision extends to a complete matter management service, emails, documents, prior work, evaluation, billing links, and strict ethical walls, all organized by client-matter. Global requirements drive multi-region storage and controls, including Australia’s residency rules. The forward look centers on differentiation through customization, firms encode expertise into models, workflows, and agents, then deliver outcomes faster and at software margins. “The value sits in your people,” Gabe says, and firms that convert know-how into systems will lead the pack.
Listen on mobile platforms: Apple Podcasts | Spotify | YouTube
[Special Thanks to Legal Technology Hub for their sponsoring this episode.]
Email: geekinreviewpodcast@gmail.com
Music: Jerry David DeCicca
Transcript
Marlene Gebauer (00:00)
Hi, I’m Marlene from the Geek in Review and I have Stephanie Wilkins from Legal Technology Hub here today to tell us about some exciting updates to the GEN-AI map that Legal Technology Hub puts out.
Stephanie Wilkins (00:11)
Yeah, thanks Marlene. It’s hard to believe, but we’re getting ready to do another quarterly update to our Gen.ai Legal Tech map. It’s been three months since we did our last one of all the solutions and Legal Tech that I’ve incorporated Gen.ai today. I know it’s like a year, right? Even in Gen.ai years, but ⁓ that was in fact at the end of June. At that point in time we saw 638 Gen.ai products.
Marlene Gebauer (00:23)
That’s such a long time, Eugenia.
Stephanie Wilkins (00:37)
placements by 505 vendors in 19 different categories, which was actually 100 more products than we saw just three months before. So to your point, three months is a very long time. Of course, since then, we’ve been working on other maps, such as our AI Agents and Legal Map and the Justice Tech Map we did with the Justice Technology Association, which we’ll continue to update. But we’re back to focusing on Gen.AI solutions. It’s time to revisit that. And now…
Every time we do one of these maps or update it, you know, we inevitably hear from people who think they should be included or say they didn’t know about it. Um, and we, course, do post some LinkedIn calling for the information and we still get those comments. So consider this your notice. If you have a legal tech offering that incorporates generative AI, we want to know about it. Cause if we don’t know, we can’t include you. It’s just that simple.
but help us help you. You can let us know what you’re up to by reaching out to us via email at curation at legaltechnologyhub.com. And if you want to see the previous maps, if you don’t know what it is, what we’re talking about, to get a sense of what we’re looking for, you can go to legaltechnologyhub.com, type in map in the search bar, click on the contents in the search results, and you’ll see a number of those graphics.
We can’t wait to hear what you’re up to and have us add you to our next map.
Marlene Gebauer (01:58)
Yeah, it’s a fantastically useful tool and everybody should go check it out. So thank you, Stephanie.
Stephanie Wilkins (02:04)
Great, thanks.
Marlene Gebauer (02:12)
Welcome to the Geek in Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gabauer. This week, we welcome Gabe Pereyra president and co-founder at Harvey. Gabe, it is great to have you on the Geek in Review.
Greg Lambert (02:19)
and I’m Greg Lambert.
Yeah, welcome.
Gabe Pereyra (02:29)
Thanks so much for having me.
Greg Lambert (02:31)
So, Gabe, you you’ve had kind of a muted background in your career. you know, just working at little places, you know, like Google DeepMind and Google Brain, you know, and you went from that to, yeah, to co-founding a legal tech startup. So, ⁓ one, you know, why, you know, why would you make a decision to go into law and think that
Marlene Gebauer (02:43)
You know
Fancy.
Greg Lambert (02:58)
that’s the right domain to apply this expertise that you learned at Google. And how did you and Winston Weinberg come up with the idea to kind of shape what Harvey is?
Gabe Pereyra (03:11)
Yeah, I started doing AI research maybe a decade before starting Harvey and so kind of 2014 right when deep learning was starting to take off and the way I ended up at DeepMind at Google Brain was kind of reaching out to
Now they’re called the godfathers of AI, but Yoshua Bengio, Jeffrey Hinton, some of the folks that kind of were the early pioneers of this stuff. And both my parents were PhDs in CS and they did kind of adjacent stuff in their careers. And so I think I was kind of always drawn to it, but I always knew I wanted to apply this technology. And so a lot of doing research at the time was really deeply understanding this technology. then kind of half my time was research and then
half
was trying to figure out where to apply this. And I was always drawn to these kind of like large societal problems. And I thought for me it was going to be ⁓ education. I left DeepMind to start kind of like a personalized education company. like to me, it felt very meaningful to like use this technology to help society in some way. I think it was kind of early at the time and there were some some challenges with doing this in education. But then when I was at Meadow working on large language models, I was kind of having the same
and thoughts of like, do we apply this technology? And Winston and I were best friends, we were roommates. He was working at O’Melvin and Myers as a litigation associate.
And I was brainstorming with some of my friends from Google Brain, of just general horizontal assistant type ideas. And he said, why don’t you build an assistant for legal? And started showing me his workflows, his legal tech. I showed him the language models. He started kind of asking some of the questions that he was working on, things like that.
And kind of right off the bat, seemed like this perfect application. And then the more we dug into it, we reached out to Sam and Brad at OpenAI. They showed us GPT-4. We got funding from them. Kind of felt like all stars aligned of like perfect co-founder, perfect industry, perfect timing with this technology. My brother also was lawyer, worked at El Malvado, now works at Harvey. And so had just like a bunch of exposure and getting to work with both of them.
Yeah, I didn’t think I would do legal, but yeah, it feels like kind of the perfect application.
Greg Lambert (05:27)
Yeah,
I learned something new this week that I didn’t know about the OpenAIs when they released the chat bot version. They were actually, I think they released it like either 3.0 or 3.5 right at the beginning there. And it was more just to kind of test it and they didn’t expect it to really kind of take off because they were already developing for in the background and we’re working with folks like you.
Gabe Pereyra (05:43)
Yeah.
Greg Lambert (05:56)
And so I think they were really surprised at how what, you know, just kind of caught fire right out of the gate. I mean, when you were talking with Winston, and one of the things that I’ve always pointed to with this type of technology is this is really kind of the first time that we’ve had technology that does well with language. Do you think that’s what was kind of the key driver for that in adoption?
Gabe Pereyra (06:21)
Yeah, I mean, I think the big intuition we had around Chachapi Tea was that really the models are the product.
And so like a lot of the startup advice we got when we were starting Harvey was kind of pick a very narrow application, know, go draft NDAs, make that work really well. And I think the thing we felt really strongly, so we got early access to GPD for about six months before ChachiPT was released. But the product we built was kind of what you see now of like talk to the model and interact with it to solve this work. And it was just so general and so powerful where we would show it to, you know, in-house lawyers,
partners, associates, and every different type of task they tried to do, the models could kind of do some percentage of that task. And so it didn’t make sense to kind of fit these models to the kind of some narrow use case. It was much more of like, how do you build this very broad horizontal tool? And then I think like to your point, it’s like when I think of how most lawyers work, like a lot of their workflows.
our email, right? Like if you’re an associate, you’re getting emailed these like tasks or prompts from a partner. You need to go use this legal tech and kind of produce some text as a response back, whether it’s like a word document or an answer to that email. And so the original idea was that it’s like, can you mimic a lot of this workflow? Can you kind of build an AI associate that that is able to make every associate kind of empowered and kind of a partner where they can manage this little army of associates?
Marlene Gebauer (07:42)
So I can’t believe that it’s only been three years. It seems like it’s been, it feels like it’s longer, right? But it’s true, it has been three years since Harvey came to market and it’s an amazing story. You’ve been able to penetrate so much of the large law firm market in so short a time.
Gabe Pereyra (07:46)
It feels like 30. Yeah.
Marlene Gebauer (08:04)
got new clients and gathered new Intel, like what were, you know, what was there that law firms, you know, the problems that law firms were having that you seemed to fill? And was that something you kind of came on the market with or have you adapted since then?
Gabe Pereyra (08:22)
Yeah, so I think we did a lot of counterintuitive things very early on. I think kind of the traditional startup experiences like go for individual users or smaller parts of the market and try to move upstream. And I think when we were starting, we had this strong intuition where if we could start working with big law firms, the largest law firms, that would be the best place to start.
And we weren’t sure if that was going to be possible. Like at the time, the models still made mistakes. There was kind of like a lot of question marks. And I think we got super fortunate meeting David Wakeling at A &O. We showed him this. was kind of a senior partner at A &O that also did a lot of their innovation work. And he was like, this is the technology I’ve been looking for. I’ve been kind of thinking about legal tech. And so we got a bunch of traction with them.
And that really validated our hypothesis of this is super useful and actually more useful for these massive law firms that have a bunch of layers of review. So the mistakes of the model get kind of caught by more senior lawyers as they review it. And then I think from there.
Kind of the core problems were, especially at the time, there was a bunch around security data privacy. Do we want to send this data to foundation models? How do we get kind of dedicated capacity and all the security problems? And then I think the part that we’ve built around is like, if you use ChatGPT or any of these basic language models to do programming, legal, any like complex knowledge work, you really quickly run into the edges, whether it’s not connected to my data, it’s not
connected to the systems I use. It doesn’t have some of the expertise that I need. And so a lot of it has been just putting this into the hands of lawyers, listening to them, figuring out, Hey, I’m working on this diligence. need this connected to my data room. I want case law. so figuring out how do we build all these pieces at the individual lawyer level? And then increasingly, as you roll this out to the entire law firms, they need things like, how do I manage all my matters, ethical walls? How do I collaborate with clients? And so a lot of
it has evolved from, you we had some intuition and then now as we’ve scaled, it’s like talking to all of our law firms and figuring out how to make them better.
Marlene Gebauer (10:33)
I think, I believe like you have had a lot of impact and helped have help firms sort of overcome some of those ethical hurdles in terms of using the product, right?
Gabe Pereyra (10:46)
Yeah, I think there’s still challenges, right? Like right when we started Harvey, there was kind of the bank letters that said, please don’t use generative AI. Now that shifted and there’s a lot of bank letters that said, please use generative AI. But I think there’s still a bunch of questions of.
You know, where can you use client data? What can you train on? What can’t you? And I think there is going to be this big opportunity to really think about like all of the expertise in these law firms that’s, you know, in the work they do in their partner’s head. How do we help firms leverage all that, put that into models, into agents, but do it in a way that protects confidentiality, privacy, privilege, maintains ethical walls. And so I think there’s a lot of like infrastructure data problems to still solve, even just organizing this data.
think it’s still a big challenge.
Greg Lambert (11:31)
Yeah, I think with the Allen and Overy being able to get in there, you jump the hurdle of the number one problem that most startups have in getting into large legal, and that is this is an industry that races for second place, and no one wants to be the first to go and try, especially new technology. And so, you know, I think
You some may say that’s lucky, but I also point out that you tend to make your own luck when it comes to these things. And so getting over that hurdle, think, really kind of catapulted the ability to get that second, third, and fiftieth firm to come in when most sales cycles for legal tech is very long.
Gabe Pereyra (12:13)
Yeah, no, I
think you’re completely right.
Greg Lambert (12:18)
So, you you were talking about the chat bot interface, but that’s, you know, kind of the beginning stages of what you were doing. And now you have, you know, Outlook add-ins, Word enhancements, the Workflow Builder, which is really interesting. And so there’s, you know, this whole attempt to do what a lot of legal tech startups want to do, which is, you know, be in the area where the lawyers are.
working into the products that they’re already working in. So a couple of questions on that. How are your users kind of leveraging these new improvements to improve their workflow? And then more importantly, because I think we’re coming into the year of show me what you’ve done for me for my bottom line.
⁓ Are you finding that there’s any metrics that you’ve been able to collect to show a real true ROI on using these tools?
Gabe Pereyra (13:17)
Yeah, so I would say that to me the biggest challenge whenever you get the use case question is the similar when people ask like, what are you using chat GPT for? And the answer is kind of like everything. But I think what like the experience we.
Greg Lambert (13:28)
I drop my email in it and
Marlene Gebauer (13:29)
Yeah.
Greg Lambert (13:32)
tell it to write a response.
Gabe Pereyra (13:34)
Yeah, I mean, I’ve
started, it’s like, just put all of our company documents into it and start thinking about like, how do you…
synthesize all of these things. it’s like, I think this and Harvey, that’s kind of the big use case of like, I didn’t work in legal, but I worked in investment banking. And if you’re working on a deal, you’re getting all these emails, you need to draft these things, you need to research, you need to go learn about this like specific company or industry that you don’t know anything about. And then the case law precedent for that. And so it’s kind of just all of this and kind of handling the complexity. But I think the big challenge is, obviously, you’re working in a bunch of different tools, your context is scattered everywhere.
And so a lot of what we’re thinking about is how do we create like a matter workspace where if you’re working on a specific deal or a litigation, you can leverage generative AI over all of that context in some like powerful and secure way. so connecting to Outlook, Word, increasingly like data rooms, DMS, case law, e-discovery platforms, things like that. want to bring this onto one place. And I think one nice metric we see in our product usage is we have pretty
strong kind of daily active users if you’re just using the Harvey assistant. But what we see is a upward trend as you start using more product surface areas to the point that if you’re using for Harvey surface areas, it’s something like 70 % daily active users. And so you’re really starting to see, OK, you can use this to do a lot of your work. And then I would say the challenge with the ROI story is what we’re starting to do with some firms is can we look at your billables and look at, for example, the work that gets written off?
And how do we start using Harvey to do all of the unrealized legal work that you’re doing? But I think where we’ve seen the really big ROI stories is increasingly with law firms and their clients, we’re working with them to build these more specialized workflows. And so, for example, we’re doing this with Oric, we’re doing this with ANO, PWC, some other law firms we can’t disclose. But…
I think a lot of it is localizing it to practice areas because I think the big challenge is when you use Harvey as a horizontal assistant, it’s hard to map this to ROI. But as we start working, for example, with private equity firms and fund formation, then what you can do is say, okay, I’m doing fund formation for like a Blackstone. They have kind of here’s their LPA process. Here’s their comment memo process. Here’s how they negotiate side letters. Can we take all of our law firms expertise about fund formation, what they know about Blackstone?
put this together into some workflow that makes that efficient. And then you can start talking about metrics like we can do this fund formation more effectively, we can get better results. And so that’s kind of how we’re starting to build some of this ROI story.
Greg Lambert (16:15)
Yeah, I’m curious on one thing with ROI and Marlene and I had a conversation earlier in the day about this is that there are some benefits that you get from tools like Harvey and others that are really hard to measure. that is things like, one is I can do something that I couldn’t have done before. And so there’s like, you know,
no benchmark for that. And then the other thing is that just the lack of frustration in just doing your daily job, there’s real value in that, but that’s not a time hours calculation. ⁓ yeah, I think that’s good ROE.
Gabe Pereyra (16:51)
Yeah.
Yeah.
Marlene Gebauer (16:58)
It’s like return on experience, you know?
Gabe Pereyra (17:00)
Yeah, so this is
actually interesting where PwC is a good example where they’ve actually managed to quantify some of this, where they’ve said that using Harvey in their early year associates is actually a really big recruiting benefit. And so they are actually able to improve recruiting because there are other places where you go and you can’t use this technology. And once you use this in your personal life, you’re like, I need to use this. And so I think there is.
to your point, it is harder to measure, but there’s these indirect benefits. But I remember early on, Winston showed me a fishbowl, which had kind of all this like big law chatter, kind of like people talking about it. And there was kind of first year for. I mean, it’s.
Greg Lambert (17:40)
Yeah, I remember Fishbowl. We actually had the Fishbowl CEO on.
Marlene Gebauer (17:41)
Me too.
Gabe Pereyra (17:44)
It’s an awesome, yeah, and there was like, there was associates saying like, I would trade my bonus for like a working laptop. And so I think this is like really a thing of just quality of life. But I mean, what we’re starting to work with firms is like, at different timescales, how do you measure these things? Because I think a lot of these things you will end up capturing in terms of like, better recruiting classes, winning more business. ⁓ And so we’re starting to think a lot and work with firms to figure out how to measure that because I do think these are kind of the important benefits.
Marlene Gebauer (17:45)
We did.
I mean, do you ever get any, any, I don’t know, push backs, not the right word, but, but, you know, people that are like nervous, like they’re like, were talking about, you know, associates and, and, know, the experience that you were sharing is they, they want this thing. And, and I’ve seen that too, but at the same time, are they worried? ⁓ you know, this is going to take my job type of thing. And, and, and, you if you hear that kind of, do you, how do you cope with that?
Gabe Pereyra (18:31)
Yeah.
Yeah, I mean, I think we got this a lot and to me this is a bigger problem beyond just legal, right? Like these models are getting very, very good and it is scary as they start doing more, like they do PhD level math, right? Like they are very good. And so, I mean, a lot of what we’re starting to talk with firms are, you know, in a couple of years when these models are, you know, maybe senior associate level, how are you training your next class of partners? How are you training associates?
And I mean, what I will say is like there will probably be some structural changes as these models get better. But I would say the opportunity is also like when I think about when I was learning programming, you know, 13 years ago, it’s like I went on Stack Overflow. I would go read like most questions. I was just like, I can’t figure this out. I guess I’m just going to not learn this. And now with these models, you can just ask them and they teach you. And it’s like now when I’m learning about legal, about private equity, it’s like this is they’re incredible. And so
We’re starting to work with firms to think about, you know, how do we build curriculums? How do we build training resources where you can take past deals and say, okay, I want to learn how to do fund formation. What is the general process? How would I negotiate this LPA? What are the parts that matter? And so I think you’ll be able to upskill people much more quickly. but I still think this is like a big challenge and something like we and law firms need to think a lot about. We also partnered with law schools. So I think that’s kind of.
another really good way to do this of giving students this technology and helping them figure out like this is the new way to practice law. ⁓
Marlene Gebauer (20:10)
They can, they can
help figure out how, what it’s going to look like. Yeah. I really love that idea.
Greg Lambert (20:15)
Yeah, had actually on LinkedIn, I had someone reach out because I had posted some things that you guys were doing. And her response was, well, you know, how do I trust that if I give it to or tell it to give me a timeline on all these documents? And, you know, other than me going in and reading every single document, you know, are you still seeing that? Yeah.
Gabe Pereyra (20:15)
Yeah.
Marlene Gebauer (20:37)
How do we do that with eDiscovery?
Gabe Pereyra (20:39)
Yeah, yeah,
no, I mean, this we’re we’re building these workflows with some litigation firms of can you build chronologies and things like this? And I mean, I think part of what is missing from these foundation models.
is that like law firm expertise or intuition where it’s like, what is a chronology? How do you build a good chronology? And so I think a lot of what we’re starting to think about is like, if you think of the expertise of these law firms, it is how do you build that work product? How do you evaluate it? And so a lot of what we need to help law firms build is how to build all of the guardrails and the ways to train these models. So if you build a chronology, you can quickly evaluate that. And I think that will be a lot of the
transformation, right? Like if you think of like
what it is really learning to be a senior associate. A big part of that is like, have a bunch of first year associates that also don’t know what a cron is and are maybe going to make mistakes. And like part of me scaling my managerial ability is like, I know which parts of their work to check. I know how to sample parts and be like, okay, this guy seems pretty burned out. He should like redo this the next day or something like that. And so it’s similar with the models, but I think the weird part is they make mistakes in like very unintuitive ways. And so part of it is just using them a lot. So you build that mental model of like,
I put too much context, it’s probably going to make a mistake here. And like a lot of that is just using the technology.
Marlene Gebauer (21:58)
So I think we could talk about this kind of stuff all day long, but, I, I am going to shift the conversation a little bit. So to talk about some of the new partnerships, ⁓ that, that Harvey has. So, latest efforts are focused on adding partnerships for, the data law firms need. so research with Lexis Nexus and Walters Clure.
Greg Lambert (22:01)
Yeah.
Marlene Gebauer (22:23)
internal data with I manage and net docs and even getting into in-house partnerships with ironclad. So what’s the goal in creating the partnerships and what do think the outcomes will be?
Gabe Pereyra (22:35)
I think the goal is obviously you want to get all the context in one place so you can leverage these models. think when we think of the type of company we want to build, you know, we don’t want to build a CLM. We don’t want to legal research data sets. Like I think the company is building these have incredible expertise and it’s kind of not our core competency. And so for us, the best world is how do we partner with them?
can we build these magical experiences where we can leverage the technology they’ve built within Harvey. In some cases, we’re providing Harvey to them so they can use Harvey in their product. And I think that kind of how do we work with this industry to build this technology together is kind of the goal there.
Greg Lambert (23:19)
⁓ Do you see like the flip side of that, for example, I know LexisNexis has come out with their own general AI tool similar to what you get started out with. So do you see kind of the cooperation, also the, what do they call it? you’re competition or something like, where you’re competing, but also kind of
Gabe Pereyra (23:42)
Yeah, and yeah, think there’ll always be
Greg Lambert (23:45)
Yeah.
Gabe Pereyra (23:47)
some of that and I think it makes the industry better where I think we can learn from each other. I think as these agents get better, maybe these agents are able to cooperate and hand things off where…
Maybe you ask a generic legal research question, Harvey, we give you some answer and you want to dig deeper and now you ask the Lexus agent or something like that. And so I think a lot of it is working together and figure out like, where are the boundaries? I mean, I think one of the challenges with this technology is this interface blurs the boundaries between all of these product surface areas. And so that is just a new product challenge as you’re building with partners and everyone. So I think something we’re figuring out with these partners together.
Greg Lambert (24:26)
Do you see more partnerships along the line?
Gabe Pereyra (24:31)
I think we are definitely looking to partner. So we kind of have partnered with case law providers, DMS providers. We want to partner with data rooms.
and eDiscovery platforms, and I think probably eventually do billing as well. And so just thinking about, you know, what are the things lawyers need to do their job well? And yeah, I think that’s the thing where it’s like you’re always missing some of this context and the more you can bring it in one place, I think that ends up being very powerful. And so figuring out the best way to do that in RV.
Marlene Gebauer (24:48)
Main data points, yeah.
Greg Lambert (25:02)
So you and I had a conversation leading up to this, which I kind of walked away from. And poor Marlene was on a, I think she had a conflict so she couldn’t sit in for the whole thing. But we ended up kind of talking about what your vision is for where Harvey is going next. And I actually looked at some of the growth that you’ve done this year and.
I went on Google Gemini and had to do some deep research on, but it was talking about the growth I think between March and July and it gave me a number of, essentially it grew by 16 and a half million dollars a day in value in that time, which is crazy. But I think we see Harvey right now as a chat bot,
a to store documents, a place to access the AI and work with it. But you’re not seeing that as kind of the four walls of where you are. And I know part of that is, we talked earlier about being able to build your own workflows. And the way that I’m seeing that is you’re kind of like lowering the barrier for law firms like ours.
to actually create products that we don’t just sell our services to the clients, but may actually end up selling the products to the clients. Is that something that you see? And then I want to follow that one up after you that answer.
Gabe Pereyra (26:36)
Yeah, I
mean, we’re already doing this. so, I mean, when I think of like, what is the…
special sauce of these law firms, it is their expertise, right? And so when we think about long term, what we want to enable for these law firms is you have thousands of the smartest lawyers in the world. How do you put all of that expertise into models, workflows, agents that you can either use to do your client matters more effectively, or increasingly, you can package those into legal solutions that you can sell to your customers, either through Harvey or
through, we’ve helped law firms build their own kind of white labeled products. And so that to me feels like the very big opportunity, we kind of just scratched the surface of this, we’ve started building these workflows with law firms and their clients in collaboration. But I think we’re so far from where this can can get to. And I think especially with how good these models will get your ability to fine tune them, customize them. So yeah, that to me seems like one of the really big opportunities here.
Greg Lambert (27:40)
And one of the things we talked about and you’ve written about is looking at Harvey as what you’re calling a complete matter management service. from beginning to end, everything that is involved in the matter is in Harvey in some way or form. Can you explain that to us, what you mean by that?
Gabe Pereyra (28:06)
Yeah, so I think if you use Harvey or most of these products today, they’re still being used in somewhat of an ad hoc matter, right? Like you will upload documents, you’ll ask some query. We have some client matter functionality, but I think when I think of the work of an associate, you’re basically living in a specific client matter, right? You get assigned some number of these and you’re working on a specific litigation or a specific transaction. And really the context you care about is what are all the emails, all of the
work product, the work my coworkers are doing on that specific client matter, and that may be precedent from similar client matters or knowledge my firm has about that client matter. And so I think the more we can centralize our product around kind of that unit. And then if you think of it as we start thinking about more than just lawyer productivity, but firm productivity, that’s also kind of the unit that law firms care about, right? They want to ask questions like, what are all the matters we’re working on? How much do we usually make from specific matters? Are we billing
as much as we should for these different matters. What is all the work that I’ve done historically for this client? Can I find all their client matters and answer questions over that? And so I think kind of unifying that both in terms of the context. So one example is it’s even pretty hard to just associate emails with a client matter. And so with some law firms, we’re starting to talk about, can we build them systems that can classify their emails to the right client matter? So you have all that context in the right place. And then eventually, you know, maybe connecting this to
billing. And then I think you also need the client matter abstraction for ethical walls. Right. I think that’s another concern as you start using these systems on like a lot of client data. It’s like, how are you, you know, giving your clients confident that you’re maintaining the ethical walls, you’re protecting that data at the end of that client matter, that data is getting deleted, for example. So I think there’s kind of like if you look at most legal tech products, like they are built around this client matter abstraction to some extent.
⁓ So I think kind of pulling all that together at that level is something we’re building into the product.
Greg Lambert (30:06)
Yeah. And one more thing that wasn’t on the list, but I saw just dropped this week is you’ve now entered into the Australian market and we’ve had Australians on before and we know that they’re like their data rules are really kind of different than the rest of the world where the data has to be there in Australia. So, you know, how are you managing the regulatory
⁓ work that requirements that now that you’re essentially worldwide
Gabe Pereyra (30:41)
Yeah, so this is actually something we started solving from the very beginning. So right off the bat, when we started working with ANO, PwC, and then US law firms, we’re already multi-region because of kind of some of the Europe requirements versus US. And so we have been building the infrastructure where law firms can store their data where they want. I think with some of these very large law firms with a PwC, you can have cases where the organization has different data requirements, different cloud requirements. And so, exactly.
Marlene Gebauer (31:08)
offices.
Gabe Pereyra (31:10)
And so like some of these very large law firms you have different countries operate separately and don’t actually share their tech and things like that. And so there is a lot as we’re doing. then when we work with large private equity firms, how do you kind of manage groups of these entities and things like that? And then all the data security, like you said, the compliance and regulatory and so something we’ve been doing for like three years now, but a challenge as you kind of go into any of these countries.
Marlene Gebauer (31:34)
So we are asking our guests now, we want to know what are areas that, where are places that you go to keep up with current events?
figure out what’s happening in the market, to learn. We want to share it with our listeners because clearly you are a visionary, the folks at Harvey are visionaries, and so we want to get in on some of that. It’s like, how do you, how do you, how do you, do you? It’s like, other than looking in your Slack, which we joked about earlier, besides that.
Greg Lambert (31:59)
Ha ha ha.
Gabe Pereyra (31:59)
Yeah, I mean, we were joking.
I was gonna say a joke here. Yeah.
Greg Lambert (32:05)
Yeah.
Gabe Pereyra (32:10)
I would say to me the big…
sources. It’s like I use ChatGPT and our web browsing to do a lot of this. I think these models are getting so good at synthesizing this content. And I think part of the big challenge, especially for a lot of this AI stuff is it’s not being published anymore. And so I think we’re super fortunate to work with all of the foundation model providers, the cloud providers. A lot of my friends from these research labs work at these places. And so you can get some intuition of
you know, what’s coming down the line in terms of AI. I think on the legal side, your guys podcast talking to our clients. But I think it really is just like talking to people. I think it is still challenging to to find a lot of this news. Yeah.
Marlene Gebauer (32:51)
Yeah, I think it’s challenging to find the news other than sort of general or, you know, applications like, okay, how, you know, how does this apply to my situation? ⁓ you know, those types of things, I think you’re right. It still requires sort of that connection with people to, you know, share like, okay, this has been my experience and, you know, this is what’s coming. This is how it might impact you. Things like that.
Gabe Pereyra (33:03)
Yeah.
Yeah, no,
and think even if you can’t talk with the right folks, it’s like starting to find the right people you can follow on their podcasts, their sub stacks, because I think it’s always hard to find, like you said, the general, but you can usually find people that are talking about the specific thing. so like when I was doing research, there was always like specific researchers where I’d read all their papers and like you could kind of follow the themes based on the work they were doing. I think that tends to work pretty well.
Greg Lambert (33:40)
And I found this is one area where you don’t have to be in legal tech to get some really good information. like this morning I just learned that Claude now does great with Excel files, that it will actually, you can input Excel files with VLOOKUPs and it will understand how it is, which.
Gabe Pereyra (34:01)
Yeah. No, mean, even
like even for us, I think the model stuff is hard to keep up with. Like there’s so much happening.
Greg Lambert (34:05)
man.
Marlene Gebauer (34:06)
So we have come to the time where we have our crystal ball question, Gabe. And this is where you’re kind of looking out into the future and telling us, you know, we usually say in the next couple of years, but maybe it should be the next few months. You know, what do you think, you know, Harvey or the legal industry needs to start addressing right now?
Gabe Pereyra (34:29)
I would say the things we’re thinking about and kind some of the interesting research directions are.
really starting to think about how you customize these models, these workflows. I think a big question we get asked a lot from law firms is, you if everyone is buying Harvey, what is our differentiation? And what I always tell them is, you know, before Gen. AI, you were kind of in the same position where everyone bought the same legal tech, but the differentiation really came from your lawyers, how you run your law firm, how you structure your practices. And I think that is going to be true in the future.
I think it’s just going to look different. And so a lot of the conversations we’re having are not just, you know, which legal vendors should you buy, but how do you upskill your associates? How do you train people to use this technology? What infrastructure should your firm be building? How should you be thinking about how you work with clients, how you, you know, price legal services, how you communicate our wide tier clients. And I think the really big opportunity here for the industry.
is I think this technology will let law firms transform not just their tech stack but their entire business. And I think you need to look at this holistically. But to me that’s kind of the big thing we’re focused on working with our law firms and their clients and I think that will be the big thing next year.
Greg Lambert (35:51)
So you mean the values in our people, not necessarily our tech?
Gabe Pereyra (35:55)
I think that’s right. Yeah, I I think like the infrastructure.
Marlene Gebauer (35:57)
We still need critical and creative thinkers.
Greg Lambert (35:59)
You well.
Gabe Pereyra (35:59)
I think the law
firms that the law firms that learn to take their expertise and put these into models and workflows, I think those are going to be the most successful law firm. Because if you think of the challenge with law firms, your revenue scales with headcount, right? Like you can only bill so many hours. But the big opportunity here is if you can take that expertise, you can put it in models. You can sell that to your clients in new and creative ways. It’s like now you can unlock, you know, exponential scale software margins. And I think there’s kind of new business models for
for law firms that figured that out.
Greg Lambert (36:32)
Well, words to think about as we move forward. So Gabe Pereyra I want to thank you very much for taking the time to talk with us here on the Geek in Review. It’s been a pleasure.
Marlene Gebauer (36:44)
Thank you.
Gabe Pereyra (36:44)
Thank you
guys so much for having me.
Marlene Gebauer (36:47)
And of course, thanks to all of you, our listeners for taking the time to listen to the Geek in Review podcast. If you enjoy the show, share it with a colleague. We’d love to hear from you, so reach out to us on LinkedIn and we’ve been posting on TikTok as well.
Greg Lambert (36:59)
Yeah.
All the kids are on TikTok, ⁓ So Gabe, for our listeners who want to learn more about what you’re doing at Harvey, where’s the best place to send this?
Marlene Gebauer (37:02)
Yeah, it’s true.
Gabe Pereyra (37:11)
I think Harvey website, the blog, and if you’re a law firm, definitely reach out to our team.
Marlene Gebauer (37:17)
And as always, the music you hear is from Jerry David DeCicca Thank you, Jerry.
Greg Lambert (37:21)
I’ve got his album up on the wall back here. There it is, back here. On the widescreen you can see it. All right, thanks everybody.
Marlene Gebauer (37:23)
Show it to go. up on the wall. All right. There we go. Promo for Jerry.
Thank you. Bye.