# Inside Harvey AI: CEO Winston Weinberg on How a 4-Year-Old Company Is Taking on the Foundation Labs

Markdown copy of https://realalpha.app/v/ekXJCDNEmOM/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Winston Weinberg says Harvey must keep reinventing its product and organization as it competes with foundation labs for legal AI.

- Show: Sourcery
- Hosts: [Molly O'Shea](https://realalpha.app/people/molly-oshea/)
- Guest: [Winston Weinberg](https://realalpha.app/builders/winston-weinberg/) is cofounder and CEO of Harvey.
- Published: Jun 16, 2026
- Length: 38 min

## The best minutes

- [17:32](https://realalpha.app/v/ekXJCDNEmOM/?t=1052) [Harvey reinvents its org every six months](https://realalpha.app/s/harvey-reinvents-its-org-every-six-months/) (4 min)
- [26:54](https://realalpha.app/v/ekXJCDNEmOM/?t=1614) [AI companies need faster decisions and relentless product work](https://realalpha.app/s/ai-companies-need-faster-decisions-and-relentless-product/) (3 min)
- [30:00](https://realalpha.app/v/ekXJCDNEmOM/?t=1800) [Vertical AI can show the return on every token](https://realalpha.app/s/vertical-ai-can-show-the-return-on-every-token/) (4 min)

## Also worth watching

- [13:00](https://realalpha.app/v/ekXJCDNEmOM/?t=780) [Harvey uses synthetic legal data to train models](https://realalpha.app/s/harvey-uses-synthetic-legal-data-to-train-models/) (5 min)
- [21:05](https://realalpha.app/v/ekXJCDNEmOM/?t=1265) [Lawyers joining Harvey must adjust to performance-based exits](https://realalpha.app/s/lawyers-joining-harvey-must-adjust-to-performance-based/) (2 min)

Skip the 10-minute office tour, the sponsor reads, and the closing conversation about Winston's routine meals.

## Full analysis: Winston Weinberg says product drove Harvey near $300 million ARR

- Harvey reports around $300 million in ARR, with token use probably reaching 12 or 13 trillion this month.
- CEO Winston Weinberg credits product work, but gives no revenue or cost breakdown.
- Buyers should connect task-level usage to outcomes before treating token growth as value.

Its 6 sections, what it means for you and what to watch are on [the page](https://realalpha.app/v/ekXJCDNEmOM/#analysis), free with an account:

- [14:27](https://realalpha.app/v/ekXJCDNEmOM/?t=867) Product use is Weinberg’s explanation for Harvey’s growth
- [15:27](https://realalpha.app/v/ekXJCDNEmOM/?t=927) Weinberg ties cloud agents to an abrupt usage jump
- [16:34](https://realalpha.app/v/ekXJCDNEmOM/?t=994) Synthetic legal data could support task-specific models
- [30:25](https://realalpha.app/v/ekXJCDNEmOM/?t=1825) Harvey’s vertical edge depends on execution
- [18:18](https://realalpha.app/v/ekXJCDNEmOM/?t=1098) Weinberg feels pressure to reorganize every six months
- [33:49](https://realalpha.app/v/ekXJCDNEmOM/?t=2029) Token ROI will test the intelligence-selling thesis

## Summary

Winston Weinberg is the CEO and cofounder of Harvey. The company’s AI platform is used by 2/3 of the AmLaw 100 and more than 500 in-house legal teams, according to the episode description.

Weinberg says Harvey’s growth has come from product improvements and a move to cloud agents, which he links to rising usage and roughly $300 million in ARR. He argues that AI-native companies must raise the product bar, make reversible decisions quickly and reorganize every six months. He sees foundation labs as Harvey’s main competition and says vertical companies need task-specific products and lower-cost models. He predicts every company will sell intelligence and argues that AI buyers will need to show the ROI of token spending, much as law firms account for billable hours.

### Minute by minute

- [0:08](https://realalpha.app/v/ekXJCDNEmOM/?t=8) **Cloud agents drove usage higher.** Winston Weinberg says token usage was 1 trillion in January and would probably reach 12 or 13 trillion that month. He says DAU/MAU rose from around 36% at the start of the year to 51% or 52%, and usage began doubling quarter over quarter after Harvey moved to cloud agents.
- [1:12](https://realalpha.app/v/ekXJCDNEmOM/?t=72) **Harvey blends legal and tech design.** Weinberg says the team used a tiny gavel as a joke when testing product latency, hammering it to urge the product to get faster. He says Harvey’s brand and office combine tech with classic legal and historical references, drawing on ideas such as the best tech companies and Succession.
- [3:14](https://realalpha.app/v/ekXJCDNEmOM/?t=194) **Local demand shapes office expansion.** Weinberg says Harvey has about 960 employees across 12 offices, with roughly 350 in San Francisco and around 300 in New York. He says the company first opened offices in response to major customers; in some countries, sensitive-data rules also required local Azure instances. Harvey also considers the legal market in each country.
- [5:12](https://realalpha.app/v/ekXJCDNEmOM/?t=312) **Lawyers and lunch build connection.** Weinberg says Harvey employs over 200 lawyers, but only about 25 do commercial legal work; others work in product, go-to-market or other roles. He says shared lunches help people across functions mix, and he avoids customer meetings during the loud lunch period, which runs from 12 to 1:30.
- [8:56](https://realalpha.app/v/ekXJCDNEmOM/?t=536) **Harvey grew through eight Airbnbs.** Weinberg names the speakeasy as his favorite room. He thinks the team went through eight Airbnbs, increasing the rental as headcount grew until they reached about 20 people. He says booking them personally was simply easier.
- [13:00](https://realalpha.app/v/ekXJCDNEmOM/?t=780) **Product improvements drove revenue growth.** Weinberg says Harvey is about four years old, has around 960 employees and 2,000 customers, and is at roughly $300 million in ARR. He says product improvements drove growth from $100 million in ARR last August toward $300 million, while the product roadmap stayed broadly consistent and build times shortened.
- [15:39](https://realalpha.app/v/ekXJCDNEmOM/?t=939) **Legal data opens model training.** Weinberg says legal post-training has been hard because relevant documents are often unavailable, so Harvey built a pipeline for synthetic datasets across legal use cases. He says 42% of customers are in-house corporates, financial services is the company’s fastest-growing vertical, and the product will need sector-specific versions.
- [17:32](https://realalpha.app/v/ekXJCDNEmOM/?t=1052) **Harvey must reinvent every six months.** Weinberg says hiring, promotions and moving people into new roles are key to scaling Harvey. He says the company needs to make major changes about every six months, and describes auditing calendars and requiring himself to write a paragraph explaining why he will take on a task.
- [21:05](https://realalpha.app/v/ekXJCDNEmOM/?t=1265) **Lawyers adapt to tech's meritocracy.** Weinberg says recruiting lawyers was difficult at first, partly because some candidates struggled to understand equity; many wanted to practice law without working in big law. He says lawyers can move into roles such as product management, but adjusting to performance-based advancement and possible dismissal in tech can be difficult.
- [23:27](https://realalpha.app/v/ekXJCDNEmOM/?t=1407) **Acquisitions should prioritize talent.** Weinberg says Harvey values strong teams more than legacy technology and has made acquihires outside legal tech. He argues that acquisitions should focus on talent and cultural fit, and warns that absorbing too many teams could destabilize a young company.
- [26:54](https://realalpha.app/v/ekXJCDNEmOM/?t=1614) **AI companies need faster decisions.** Weinberg says founders should treat most decisions as two-way doors, make calls at about 51% certainty and correct course quickly. He argues that enterprise AI has a much higher product bar than SaaS did because users can turn to other products, so vendors must keep improving.
- [28:34](https://realalpha.app/v/ekXJCDNEmOM/?t=1714) **Companies will sell intelligence.** Weinberg says every company will eventually sell intelligence. He argues that Harvey needs task-specific models that perform as well as or better than frontier models at significantly lower cost, potentially a hundred times less, and says companies will sell both a product layer and intelligence.
- [30:00](https://realalpha.app/v/ekXJCDNEmOM/?t=1800) **Foundation labs are Harvey's main competition.** Weinberg says Harvey’s product and models focus on specific legal tasks, which he thinks will be difficult for the labs to match at the same level of specificity. He identifies the labs, rather than legal-tech companies, as Harvey’s main competition and says successful verticals will keep drawing their attention. He says Harvey’s win rate in Europe is over 70%.
- [32:52](https://realalpha.app/v/ekXJCDNEmOM/?t=1972) **AI buyers need task-level ROI.** Weinberg argues that buyers should ask whether every task needs frontier intelligence and how they will measure the ROI on token spending. He compares this to law firms’ six-minute billable-hour entries and says vertical vendors can connect token use to value for specific tasks. He also says legal benchmarks often test bar-exam multiple choice rather than end-to-end legal work.

## Transcript

Sourcery, hosted by Molly O'Shea, with Winston Weinberg.

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[0:00](https://realalpha.app/v/ekXJCDNEmOM/?t=0) The purpose of Harvey, Harvey AI, Harvey AI, if you're reviewing a million documents, is how do you check that its work is accurate? So you were at $100 million in ARR last August, and now you're nearing $300 million. What's been driving that? It's 100% product. Our token usage in January was $1 trillion, and this month it'll probably be like $12 or $13 trillion. Our DAU over MAU at the beginning of the year was around 36%. Right now it's like 51%, 52%. And then this year, most recently, the main switch we did is just went over to cloud agents. We switched our entire infrastructure to that. And once we did that, usage user just started doubling quarter to quarter.

[0:46](https://realalpha.app/v/ekXJCDNEmOM/?t=46) You asked about a sword. We could start with this. We could start with this. So I actually, this is probably the best thing. I also personally have a replica of Frostmourne. uh but that's in my apartment i don't know what that is it's it's arthas's sword uh world warcraft got it yeah anyway of course my bad my bad it's an awesome sword but i've seen too many people do sword things and so i didn't end up doing it um but i do have it it's great you guys can okay well that works too you can come through this way it's totally fine um but yeah this thing's fantastic we've actually we had a really small one um there was a kind of like a running joke when we were all in an airbnb that we had this like little tiny gavel and one of the biggest problems with the product was just like latency um and so whenever we tested like a new feature we would basically just like hammer the gavel and be like make it faster make it faster make it faster which i think annoyed that a lot of people but we've stopped doing that now oh my gosh wow okay that's a good place to start yeah yeah we got a bunch of books got some sculptures and you handpicked all of this stuff just for reference we're on a tour of harvey right now

[1:53](https://realalpha.app/v/ekXJCDNEmOM/?t=113) this is Winston and um what are we gonna see today um we're gonna see the floor that's been renovated so three floors here in San Francisco and then one the one we're on now has been like fully renovated okay and then we're doing the sixth and the seventh and the significance of this painting yeah so there's no significance of this painting but um I mean I think like one thing about this floor in general is we've tried and we'll take you up to the speakeasy after we've tried to do like a combination of kind of like tech plus also like classic um and so you'll see like a lot of sculptures we have a lot of uh like books related to law this is kind of funny because it's John Grisham but it looks like it's like oh my god ancient text like Hammurabi's code or something like that how did you find them um I actually don't know for these particular ones but in general we've kind of tried to do like a combination of um when we were actually doing our brand in the beginning we were basically like could we do a copy of like the best tech companies and then succession like the show yeah my point was like kind of a combination of like an ode to you know like classics like greek first orators things like that plus like hey we're a tech

[3:08](https://realalpha.app/v/ekXJCDNEmOM/?t=188) company um and we've definitely tried to do that a lot with our brand and we've tried to do that throughout the office too. It seems to be working. It looks amazing. Thanks. Okay, so we're going to cover a lot of different things today. For starters, on this tour, I want to talk about the evolution of the company. So you've been around close to four years now. Yeah, it'll be four years in August. Congratulations. Well, not yet. Happy birthday. I don't know. You have nearly a thousand employees yeah close it's like 960 something and this office is one of how many uh 12. wow so we opened up offices pretty quickly but the the office here and then new york are by far the largest how many employees are um so in this one i think it's around like 350 and then New York is there.

[3:59](https://realalpha.app/v/ekXJCDNEmOM/?t=239) around like 300, something like that. Yeah. And then the vast majority of EPD is here, but we are growing out the other offices too. That's a really cool vibe. Yeah. Okay, so when you build out globally, you have 12. Yeah. How do you figure out which cities to go into first? Is it like you have your customers and then you build around them? Yeah, so it's actually, yeah. So we did it first based off of just like reacting to like big customers. So like we'll sign like Deutsche Telekom and it's like, oh, wow, we need an office in Germany, right? Things like that. There's actually something really interesting that we had a problem of in the beginning, which is because we process sensitive data, a lot of the countries, we actually needed like an Azure instance in each one, right?

[4:44](https://realalpha.app/v/ekXJCDNEmOM/?t=284) So like in Australia, you can't process financial data outside of the country. And so we would set up these offices and then we'd set up like Azure instances too. And it was almost like we set up an Azure instance and then that would be like a pretty good indicator that we'd have to set up an office pretty soon afterwards just because like customer demand yeah um but another way to look at it is kind of like the legal tam overall too um and so we kind of base it off of like you know how many lawyers are there in in each country etc and there's how many lawyers in this company yeah um in this company we have over 200 yeah and so and then on the commercial side like actually i guess doing what a normal lawyer would do um there's only about like 25 really yeah and then the rest um i mean obviously we use harvey internally for a lot of stuff so we're trying to scale that up um and then the rest of folks either they work on product or they like help with go to market and things like that too um it was we don't have this anymore but you used to if you would have visited us like a year or two ago you would have been able to tell who's a lawyer and who isn't because why there was they would dress differently yeah so there

[5:56](https://realalpha.app/v/ekXJCDNEmOM/?t=356) was like a phase should we play a game right now yeah you won't uh you won't be able to actually i don't think anywhere on this floor actually um but basically like the first week they would wear a full suit and a tie and then about like maybe the next monday the tie would come off and then like three days after that it's like the suit came off and then it would eventually be like hoodie and sweatpants yeah yeah naturally and you but you would be able to tell like basically like which lawyer had been here for which amount of time and you could always tell which is kind of funny so as we walk through this like how did you set up this new office like there's obviously a lot of conference spaces yeah desks this is a huge communal area do you guys host lots of yeah we we host a lot of hosts like customers and things like that we also i mean it's a little bit past lunch but we were talking about this earlier we have like an insanely loud lunch culture yeah we walked in at like 12 30 and i was like yeah this might be it's it's crazy loud like to the point where i don't schedule meetings during the lunch really like i or i don't do customer meetings during the lunch hour like it's literally like a band like from 12 to 1 30 i just will not do

[7:07](https://realalpha.app/v/ekXJCDNEmOM/?t=427) external meetings because it is so loud and you can't it's loud everywhere so people will eat lunch here but it's like loud throughout the whole thing um i think we just have always had that we started in airbnb and then whenever we moved from office to office like we'd always eat lunch together and it's somehow like spread out to the you know hundreds uh almost a thousand employees and so we want to keep that i think forever i don't know i think like you need like a couple of those instances where like everyone gets together we've done a handful of these office story so far and you might have the most employees in office. Really? Yeah. Well, don't tell them that. I mean, I think like it's something we care about a lot.

[7:51](https://realalpha.app/v/ekXJCDNEmOM/?t=471) I think that if you have to try to force it, you're probably in trouble, right? Like it's not gonna work that well. But I think like the thing that does it the most is stuff like This is like, you get to meet everybody from like, if you go to here during lunch, and again, you wouldn't have been able to hear anything. So maybe not as helpful. But you want you can't be like, Oh, that's the team that works on this. That's the team that works on it's like everyone is all mixed together. And so I think people like coming into the office because we have a very like culture of everyone working together. And people are really good friends that are like a cross functions. So we have people like, like, you can come here on the weekend, and there will be people in the weekend.

[8:30](https://realalpha.app/v/ekXJCDNEmOM/?t=510) really yeah all the time yeah yeah and it's it's funny because no one works in offices and so people usually work in like a corner um and then you'll have like just a small group of people in a corner but like i notoriously lose like my card 24 7. i think i have my with me right now but this is like my 19th like i just constantly go through them and so i know that there's people on the weekend all the time because i'll come on the weekend and i'll post into sf general like like, yo, let me in, yeah, and there's no one to. Maybe they think that's a test. Exactly, and so every single time so far I've been able to get in, yeah.

[9:03](https://realalpha.app/v/ekXJCDNEmOM/?t=543) Okay, so what's your favorite room? Oh, yeah, good question. Okay, favorite room is definitely a speakeasy, and I'll show you that in a second. I'd say my second favorite is I really like to work actually on the couch, like on the couch when people come in, and I think part of that is I used to travel a lot, I travel a little bit less now, but last year i almost entirely didn't work out of an office i just worked on the couch and it's because i get here early and i like like saying hi to everyone for a couple hours and then i have meetings um but in terms of actual room is the speakeasy so i can show you that it's up here actually oh cool i'm surprised by being in office i'll take that though it's uh i mean it was a good compliment yeah no no um i'm trying to think like the only other company that has was applied intuition oh yeah yeah they were packed they're young though they're like 40 people right oh oh no applied intuition not applied compute not applied compute i like applied computer gabe was just there like last week okay so what's the biggest lore here biggest lore of this office yeah this office actually doesn't have the airbnbs have like way more oh i would imagine yeah it was i mean so airbnbs we went through i think eight so basically what we did

[10:22](https://realalpha.app/v/ekXJCDNEmOM/?t=622) is at what point did you realize you needed to get 20 people okay so we what happened is we basically kept getting like a larger airbnb each month when we just added more people um and then we got to a point where it was like okay this is definitely a problem um at like 20 ish what is the deal with people in sf like startups getting like townhouses and houses to build yeah out of i would like to say that there's like some beautiful reason for this or some smart thing to be 100 honest it's just easier really it's just like i can go personally book an airbnb okay and then you just don't think about it because you're doing so many other things legal what to yes it is legal to do that yeah um and then you i mean you basically just like personally book a new airbnb each time and it's like then you don't have to worry about what's the most illegal thing that you know i'm not gonna say that on camera nothing we're a bunch of lawyers are you kidding me okay cool sorcery is brought to you by brex the financial stack trusted by more than 30 000 companies, including one in three venture-backed startups in the US. Nearly 40% of startups bail because they run out of cash. Rex is literally built to help founders avoid that. Unlike

[11:45](https://realalpha.app/v/ekXJCDNEmOM/?t=705) traditional banks that let your money sit idle, chipping away at it with fees, Rex is designed to help you spend smarter and move faster. Their all-in-one solution combines checking, treasury, and FDIC protection into one powerful account. You can send and receive money globally at lightning speeds. Get 20 times the standard FDIC coverage through their partner banks and even high yield from day one. But same day and even same hour liquidity. Access your funds anytime. Companies like Scale AI, DoorDash, Service Titan, HIMS, Anthropic, Flexport, Robinhood, and Plaid trust and use Brex. Start today at brex.com slash sorcery. That's B-R-E-X dot com slash sorcery.

[12:30](https://realalpha.app/v/ekXJCDNEmOM/?t=750) sorcery turing is training the next generation of ai with tasks that require real expertise and real world judgment that's why companies like NVIDIA anthropic salesforce and gemini partner with turing turing builds realistic reinforcement learning environments and data systems based on real operational traces the kind of infrastructure frontier labs need to train superintelligence. Visit turing.com slash S-O-U-R-C-E-R-Y. We just did a little walking tour of the office. Now we're going to go deep into the business. Are you ready? Yeah. Let's do it. Okay. So where is Harvey at today? Give me the full stats. Yeah. So we're about four years old, August. That'll be the birthday. I think it is birthday. What's the day? August 4th. It's a Leo.

[13:18](https://realalpha.app/v/ekXJCDNEmOM/?t=798) Holy crap. Wow. Actually, I didn't think about it that way. Yeah. Huh. Well, that puts like everything into perspective for me i gotta like take 20 minutes to have like deep thoughts on that i'm a capricorn so i don't know if that works is that a problem i have no idea i don't either i have no idea leo's like a lion right yeah fire sign fire um so yeah about four years old um and uh we're around 900 uh folks uh 950 people 960 something like that um we're at around 2 000 customers um somewhere around 300 million ARR um so we're going really fast um I think like it's it's mostly just a crazy experience like how much the product has grown um I think that in the beginning like we had this product roadmap and our product roadmap has actually like really been pretty consistent the main thing that's collapsed is just like how long it takes to build these things um and so I think the the thing that's actually moved the fastest is how much we are starting to basically build a platform for all of these different pieces of the product so you were at 100 million in ARR last August and now you're nearing 300 million what's been driving that uh it's 100 product I mean like I'll give you some metrics um our token usage in January was 1

[14:34](https://realalpha.app/v/ekXJCDNEmOM/?t=874) trillion like for the month of January and this month it'll probably be like 12 or 13 trillion oh my god yeah so the usage is getting pretty insane our DAU over MAU at the beginning of the year was around like 36% right now it's like 51 52% our queries per user have basically been doubling up quarter over quarter hours spent is doubling quarter over quarter I think that's better than queries the problem with queries is as you build a more complex product you're gonna have these outputs that are like crazy long and so you might have people like using this like querying less but the output is like a hundred page document and so they're like in your system reviewing it and collaborating on it um but um like everything is just product improvements i think like last year last year we had some moments where we just kind of had to like rebuild a lot of our product um and then this year most recently the main switch we did is just one order cloud agents like we switched our entire infrastructure to that and once we did that usage literally just started like doubling quarter over quarter and you've raised over a billion dollars now so have you used most of that capital whereas are you using it on tokens like what is yeah yeah

[15:44](https://realalpha.app/v/ekXJCDNEmOM/?t=944) now we're using the office now we're using on tokens um no we actually we haven't used a lot of that money um i think like the the interesting thing that we always wanted to do was actually do a lot of post training on the models and there's a couple problems in legal that makes us like really hard One is the data isn't available. So like if you went online, you're like, I want to go find a bunch of documents that are related to like a random fund formation by Blackstone. They don't, they don't exist. Like you'd have to go to Blackstone for those documents. Right. But the thing that happened with like the last generation of coding models is you can actually take sets of documents and create synthetic docs that are so good that the lawyers can't tell the difference between whether they're created you know by an actual lawyer or

[16:30](https://realalpha.app/v/ekXJCDNEmOM/?t=990) **Winston Weinberg:** they're created by the coding models and so with that we've now actually created basically like a pipeline for creating synthetic data sets across like every single legal use case and because we have that now now you can start

[16:42](https://realalpha.app/v/ekXJCDNEmOM/?t=1002) actually plus training models and that's me expensive Wow and so what are the main categories of your customers now I think you said it's 42% is in-house corporates yeah 42% is in-house corporates it's growing faster than the law firms technically um because i think we're we're at like almost 70 of the is it easier for them to adopt is that why no they're just slower to adopt yeah so the corporates adopted like they started like a year after the law firms um and so and then our fastest growing of like the of the verticals in general is uh financial services is number one for sure so like banks private equity asset management um and then pharma is actually growing pretty fast too uh and what we're starting to do is actually like we're going to verticalize our product too where we have to actually like have different parts of harvey that are different for each vertical right because like the compliance and legal needs of a bank are very very different than even private equity so i want to go into deeper building the company out today like what does it take to build an ai forward company so i reached out to pat grady and he said the one thing that he emphasized was that you've been able to

[17:50](https://realalpha.app/v/ekXJCDNEmOM/?t=1070) reinvent the company over and over again. So what was it like from the beginning to now? And what do you think was like, what are the key unlocks along the way that you think are critical? Yeah. I mean, the key unlock is hiring for sure. Like that is it. Um, and when I say hiring, I don't just mean hiring new people. I mean like also development, right. And like promoting people, putting them into different roles, things like that. Or, or honestly, in some instances, they get outscaled and like figuring out a better role for them. But every six months, i'd say i start to get like this weird feeling of like things are just breaking right um and i feel it's like a pressure that like builds up um and then usually hope so far what has happened is every time that has happened there's been like three big changes that i need to make and i realize i need to make them and then i make them and then like the pressure like flows off um and that happens like literally every six months i'd say um maybe three to six months something like that

[18:48](https://realalpha.app/v/ekXJCDNEmOM/?t=1128) **Winston Weinberg:** and it it feels like if you do not constantly change you are just going to get so behind that you die as a company right now um and i think a lot of that is actually like who do you decide

[19:02](https://realalpha.app/v/ekXJCDNEmOM/?t=1142) to promote who do you decide to hire uh that is more important than anything else because a lot of people can't do that like it is really hard for every six months for someone to like massively change how they operate um it's even simple stuff like when i see somebody like mass starting to really break the first thing i do is i'll go and we'll do like a calendar audit of just like really yeah seriously that's like the first thing it's i know it's so stupid and it's so simple but you go in and you do a calendar audit and you come up with like really good ways to be like do you have to do any of these things but tell me what is the main prior this week are any of these related to that prior and you will start seeing people and i do it myself like you just come up with so many excuses for why you're doing stuff and at the end of the day it's like no that's not relevant that's not relevant that's not relevant um and i think that like finding people that are really good at doing that themselves and i think there's a decent amount of people at this company now that have learned how to do that that's how you scale like that is how you scale it It's like crazy rapid pace that normally, you know, companies need to like reinvent themselves

[20:04](https://realalpha.app/v/ekXJCDNEmOM/?t=1204) maybe every like five years, 10 years, something like that. I think you have to do it every six months. So what's your framework for prioritization? How do you determine if you can't do something or not? Yeah. So I mean, for me, I have like a bunch of different rules. One of the really simple rules is like my chief of staff, whenever she sends me something, I have to write a paragraph for why I do it. Not kidding. You don't use an AI for this? No. It's really important because if you start for the stuff that is really, really, really important, and again, this is outside of ordinary. I don't do this for product reviews that we do weekly. If you sit down and you can write an entire paragraph about why you're going to do something, you definitely are going to do it.

[20:50](https://realalpha.app/v/ekXJCDNEmOM/?t=1250) If what happens is you go, it's so annoying that I have to write this down. Why would I ever need to do that you probably don't need to do it it's not that important right and then what i've found is the stuff that is really important i could write like 100 pages about what i'm doing right and i know it's it's like kind of a small silly thing but it helps a lot to get a massive unlock for just blocking things so you have a quarter of your company as previous lawyers former lawyers how did you convince them to come to a tech company that's completely different than a traditional career path. It was really, really hard in the beginning. And now it's like, I think quite easy. You show them equity. What, what works? Yeah. Oh my God. It was most, it was difficult to explain equity in some instances, right? Like what is the value of equity and things like that?

[21:36](https://realalpha.app/v/ekXJCDNEmOM/?t=1296) I think really like we hired in the beginning, we hired a lot of people who I think really wanted to like, they love law, like they love the practice of law, but they didn't love working in big law. and so they wanted to do something adjacent and I think that was really really attractive and then over time like we have so many lawyers that are now just full-time pms like literally they just like transferred in other pms we have so many different like career opportunities that I think that now it's a really attractive place to be because you could have a legal background and then you end up doing something else right like you don't actually have to be practicing law or anything like that here what's the biggest complaint of people leaving law firms oh i'll tell you by far they uh they like the biggest thing that they need to get used to here is and maybe that's your question if it isn't i'm gonna answer that one anyway um

[22:28](https://realalpha.app/v/ekXJCDNEmOM/?t=1348) **Winston Weinberg:** because i think this is actually super interesting law firms very rarely fire people very rarely okay um and so actually i think one of the most interesting things to get used to

[22:40](https://realalpha.app/v/ekXJCDNEmOM/?t=1360) you is that tech is very fast moving and I try to make it a meritocracy and so if you aren't doing well, we're going to have to let you go. And I think that's really hard for some people that have been in a situation where you get promoted every single year and that's how it works in big law, you get promoted every single year to exactly the same level as everyone else despite your performance and you don't really get pushed out. right and i actually think that's one of the the weirdest things and i remember in the beginning we like there were a couple people that we had to let go and things like that and people freaked out um and it's like you know from the tech world that's like very normal but i remember that being like a huge issue in the beginning thinking of how fast you're growing the company i don't know maybe you still have a billion dollars in the bank how are you thinking about buying or sorry building out the company itself or buying there's like a now there's like an onslaught of M&A within all these AI companies and buying up smaller startups for talent and all this kind of increased competition how are you thinking about that yeah devalue like higher value on team and lower value on what they've built in terms of

[24:00](https://realalpha.app/v/ekXJCDNEmOM/?t=1440) Like, I do not believe that it is a good idea right now to go around and buy legacy technology. Like, I don't. I think it is a much better idea to buy, like, really, really good teams, regardless of if they worked in your space. Doesn't matter, right? And so that's, like, if you look at the acquihires that we have done, they actually haven't been in the legal AI space or, like, legal tech. They've been outside of it. But they're really, really good teams that could work on a problem that we have, right? and that doesn't mean that I won't do legal tech acquisitions in the future but I do think that right now if you are making an acquisition like the number one thing you should be looking at is just talent because you can build things so much faster now right that it should literally just be talent like are you buying a team that is really really good and are they going to align with your cultures because the other problem is like we're not even four years old if we go and absorb a bunch of teams, like our culture is still being built, right? And so you're going to just like collapse yourself, right? And so I think it's a really bad idea to go out and buy like a bunch of companies and then they end up being like half of your company is that. That's a

[25:07](https://realalpha.app/v/ekXJCDNEmOM/?t=1507) terrible idea. I do think it is a good idea to go out there and like acquire some companies for really good talent. Today's episode is sponsored by VCX by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. Learn more at getvcx.com. Some of you may not have heard this yet, but our sponsor Public just launched something called Generated Assets, and it brings AI into investing in a way I've honestly never seen before. Here's how it works. You type in an idea like AI-powered supply chain companies with positive free cash flow or defense tech companies growing revenue over 25% year over year. Public's AI then dispatches a swarm of agents that scan every single US stock, evaluates them, and instantly builds a custom index around your thesis. What really stands out is how clearly it explains why each stock is included. And before you invest, you can even backtest your idea against the S&P 500, so you're making decisions with real context, not just guessing. And beyond generated assets, Public lets you invest in stocks, bonds, options, crypto, all in one place. They'll even give you an uncapped 1% match when you transfer your investments

[26:17](https://realalpha.app/v/ekXJCDNEmOM/?t=1577) over from another platform. If you want to build a portfolio that actually reflects your thesis, visit public.com slash sorcery. Paid for by public investing. Full disclosures in the description. Enterprise AI runs on Merge, the AI infra platform for integrations, agent tooling, and model orchestration. So your teams ship product, not plumbing. Mistral, Dropbox, and Drada already trust Merge in production. Start building at merge.dev. Founders scale faster on Deel. Set up payroll for any country in minutes. Hire anyone anywhere. Get visas handled fast and get back to building. Visit deal.com slash sorcery.

[26:54](https://realalpha.app/v/ekXJCDNEmOM/?t=1614) That's D-E-E-L dot com slash sorcery. So you started this company post-COVID. It's nearly four years old. And it was AI native from the start. So I just want to know, like, what is fundamentally different about building a company? today in the AI era versus the SaaS cycle? Yeah, I mean, I think pace is really important. And one thing with pace that I think people struggle with

[27:23](https://realalpha.app/v/ekXJCDNEmOM/?t=1643) **Winston Weinberg:** is like, you have to just assume that most things are two way doors.

[27:28](https://realalpha.app/v/ekXJCDNEmOM/?t=1648) And people really hate this. Like they really like to kind of like think about a decision, get to like 90% certainty and then make a decision. The reality is like, you're probably gonna have to make a bunch of decisions at like 51% certainty, which means you're being wrong, right? And can you deal with being wrong and then quickly pivoting, right? And I think you have to do that way more than you used to in the past, A. B, I think there's a huge difference on enterprise, like massive difference on enterprise, which is You cannot get away with like going in a room, building some product, and then selling it, and then never improving the product over the next like XYZ years. The users matter so much. Like the alternative is Quad, ChatGPT, these other things, like those are great products too. And so I think that the product bar for enterprise is astronomically higher than it used to be, where like you 100% have to be constantly innovating.

[28:30](https://realalpha.app/v/ekXJCDNEmOM/?t=1710) and creating the best product for the end user. And I think in the past, you could kind of get away with doing some things like really long sales cycles and things like that. And token usage is compressing margins. So how are you financing 13 trillion tokens? Yeah, I mean, so there's a lot of... Okay, a couple of things. One, I think every single company is going to sell intelligence. Like that is gonna be the core of the company. and so like eventually like we were selling legal intelligence we're selling legal intelligence to lawyers right but it's still like selling intelligence and so what I mean by this is like we need to get to a point where we have models that perform as well if not better than the frontier models but cost significantly less like I basically think of this as like an intelligence allocation problem where if you look at like a model you know the new models coming out it's like mythos like 5.6 all these guys right those models come out and they're incredibly expensive and they're general.

[29:29](https://realalpha.app/v/ekXJCDNEmOM/?t=1769) And if you use them across every single legal task, like maybe they do decently well, although on our benchmarks, like a lot of them fall apart, but they're insanely expensive. And so the better thing is, can you actually create models that like do diligence, do change of control review, do contracting, do these different things at the level, if not, or better than the frontier models, and they cost a hundred times less, right? And so I think that every single company is going to sell tokens. and I think that they are basically going to be selling intelligence. So you'll have like the product layer and everything like that, jubble up your data, and then you're going to be selling intelligence as well.

[30:04](https://realalpha.app/v/ekXJCDNEmOM/?t=1804) How do you compete with those large models that are offering competitive services? Like what is the thing about Harvey that you cannot copy? Yeah, I mean, the best way I think about this is like there's the product side and then there's the intelligence or model side. And you could say on the product side, it's like the harness side. I think that's like conflating things a little bit too much. on the product side you just go very vertical right so you build solutions that are really good at like diligence in a particular space like all of those things and i think it's gonna be hard for the labs to get to that level of specificity on the product side on the model side you actually do a similar thing which is you basically build a bunch of models that are really good at specific legal tasks and then you optimize them for cost right like there's a world in which gpd 10 is more expensive than a lawyer there's a that's like a very possible world right and so where we have is basically you can think of like the frontier models here commoditized models here I think a lot of the economy is actually in between these things right because the frontier model might be like too expensive to basically put in terms of every single piece of work and so you

[31:14](https://realalpha.app/v/ekXJCDNEmOM/?t=1874) have this massive space that is like all these vertical companies in terms of other competitors lagora is fast moving behind you how do you think about i mean you are global from the beginning but they are dominating europe so how do you close that gap yeah i mean i would say in europe i think our win rate is like over 70 um so i don't think it's as much dominant there to be honest i think like our main competitor is the labs i mean you saw clock for legal um there's a rumor that like codex for legal right i think that the reality is like the labs have an incredible amount of resources right and so it is a race for how quickly can we build the best product that is verticalized and then how do we build the best vertical models right and I really do think at the end of the day it is a race against the labs and I think Every single company on earth is competing against them.

[32:02](https://realalpha.app/v/ekXJCDNEmOM/?t=1922) Do you think they're going to start acquiring? Like in legal specifically? I think that they will start acquiring in any space where they see a significant amount of traction. Have they tried to acquire you? One thing that I think is crazy. I'm not going to answer that. One thing that I think is that people forget about is the more success that any of these verticals have they'll just keep entering them again and they'll re-enter them and so i think like people you know clock for legal came out recently and i think a lot of people were like oh okay well it was this release and now that's kind of it or whatever the better we do the better other companies do the more the labs will be like oh okay we're gonna put more resources into that and so it's not like a one and done thing it is like a constant competition what do you think the biggest question right now is that people are not asking by far i think the biggest thing that people are not paying attention to is what i was talking about earlier which is do you need frontier intelligence for every single task i mean this is the best example of this is like what happened with uber recently right and it's like we are going to get to this point

[33:10](https://realalpha.app/v/ekXJCDNEmOM/?t=1990) this is actually a weird situation where um i'm sure you know what the billable hour is because people like talk about that a lot right yeah weirdly the billable hour problem is the same problem that i think the entire world is about to run into like the billable hour for for those that don't know what it is. It basically is when you get a bill from a law firm, it says in six minute increments what people did and then the hourly rate or the six minute increment for that task, right? Why do they do that? It is because they're trying to show ROI. They're like, hey, your legal bill is a hundred grand. We're going to break it down into six minute increments, right? This is

[33:49](https://realalpha.app/v/ekXJCDNEmOM/?t=2029) **Winston Weinberg:** kind of like the main problem that I think the entire world is about to hit, which is I just spent a billion dollars on tokens where's my ROI right and I actually think that like there aren't

[34:03](https://realalpha.app/v/ekXJCDNEmOM/?t=2043) enough companies that I know of that are starting to think about how do I actually show ROI in every single vertical use of these things and I think that vertical companies are gonna have a huge advantage here where you can start to get to the point where you basically can show every single token and what the ROI was of that token for your particular task in vertical so before we started recording I asked you what you're most excited about and you said benchmarks yeah we're going to be interviewing Gabe after this what question do you think I should ask him um I think that the best question actually is why are benchmark like why are why are the current benchmarks for most verticals bad right um and if you look at like the benchmarks that have existed for legal for a long time i mean half of them are like can it pass the bar multiple choice questions on like community property law and things like that right and i think that we haven't actually until now had here is a very like very good set of data that doesn't a legal task from end to end and we're missing this in most articles other than coding basically coding is the only one that has like a good saturated benchmark wow okay when are you going to do a collaboration with kim kardashian

[35:17](https://realalpha.app/v/ekXJCDNEmOM/?t=2117) soon she didn't pass she gave up i think kanye passed it or maybe that was a joke i don't know i think it was a joke i think it was a joke damn i think it was a joke i think it was i'm not sure well thank you so much yeah of course thank you because it lives why do you have a hippo well because they both live on land and in the water so they see both sides of the world oh my god that kind of works ducks fly and then ducks fly ducks do and then hippos eat people hippos have eaten a few people so that's not i don't know why you want runaway jury that's legal john grisham how many i've noticed you have over 1200 books in this office have you read all of them every single one of them From front to back.

[36:00](https://realalpha.app/v/ekXJCDNEmOM/?t=2160) And I could recite any part of the book. Even the culinary cookbooks? Mostly those, actually. I'm, like, more well-versed than that. No, it's been a long time since I cooked. I am really bad. I'm a chicken and rice DoorDash guy. And it's actually the same. It's the place called Cholito Linda. And it's around here. And there was actually a time I was interviewing someone. And I asked them, like, hey, do you want lunch? And I pulled out my phone to do DoorDash. And they'd seen that I ordered it 467 times. yeah it said like the like the little number you ordered multiple times a day yeah twice lunch and dinner oh yeah so you're one of those types you just eat the same thing every day well so and then breakfast um there's a blue stone on like fulsome and i get there exactly when they open like it's 701 and i get um it's like this keen green smoothie that's really good with extra almond butter and i look like a psychopath because i'm like i have a spoon and i'm literally eating you eat it with a spoon yeah it's kind of terrifying i eat so i try to these with forks so you're okay how do you even do that wait wait actually i'm literally not even kidding how well i like to put like granola on top so i eat it with a fork but then you so is it

[37:11](https://realalpha.app/v/ekXJCDNEmOM/?t=2231) like you fork out the granola so it's like granola doused in smoothie okay so it's granola doused in smoothie and you do that with the fork and then you put a straw in it's like a smoothie parfait yeah that actually kind of makes sense you know that's not that bad a spoon seems like it would work better though. But I'm making you feel less bad about the whole spoon. I guess that makes sense. All right, are we walking or what are we doing here? Yeah, let's start. Okay, so. I like the duck though. We do. The duck was great. I mean, might as well. Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.vc, where we deliver a once a week top deals and tech headlines email and also go deeper on our podcast interviews. Subscribe to Sorcery today. And don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.