# Harvey is moving from seat sales to selling legal work

Markdown copy of https://realalpha.app/s/harvey-is-moving-from-seat-sales-to-selling-legal-work/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Winston Weinberg](https://realalpha.app/builders/winston-weinberg/) · Winston Weinberg is cofounder and CEO of Harvey. · Co-founded and leads Harvey, after working as an associate in securities and antitrust litigation at O'Melveny.
- Host: [Sonya Huang](https://realalpha.app/people/sonya-huang/) · Partner, Sequoia Capital · Works with OpenAI and Anthropic; co-wrote Sequoia's essay on generative AI.
- Host: [Pat Grady](https://realalpha.app/people/pat-grady/) · Co-steward, Sequoia Capital · Joined Sequoia in 2007; backed Snowflake, Zoom, ServiceNow and OpenAI.
- Show: Training Data
- Episode: [How AI Breakout Harvey is Transforming Legal Services, with CEO Winston Weinberg](https://realalpha.app/v/eXK-_yyQDMM/), Mar 11, 2025
- Moment: 16:01 to 21:33 of the episode (6 min), playing on https://realalpha.app/s/harvey-is-moving-from-seat-sales-to-selling-legal-work/; the episode from there: https://realalpha.app/v/eXK-_yyQDMM/?t=961

## Quote

> “And so we are transitioning from just a seat-based company to actually **selling the work** as well.”
> — [Winston Weinberg](https://realalpha.app/builders/winston-weinberg/) at [17:05](https://realalpha.app/v/eXK-_yyQDMM/?t=1025) · Training Data

## Takeaways

1. Harvey combines its technology with law firms’ domain expertise.
2. Revenue splits give firms a stake in AI-enabled services.
3. Joint projects test whether a workflow is repeatable and buyers want it.

## About this segment

Winston Weinberg describes pairing Harvey’s technology with law firms’ expertise and selling AI-enabled workflows to clients. Revenue splits help firms earn from work that was once done at a loss.

From “How AI Breakout Harvey is Transforming Legal Services, with CEO Winston Weinberg”, Mar 11, 2025. [Watch the full episode](https://realalpha.app/v/eXK-_yyQDMM/)

## Transcript

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

[16:03](https://realalpha.app/v/eXK-_yyQDMM/?t=963) Is this legal software or is this AI lawyers? Yeah. So, so far, I mean, software. I think the best way, like most of our clients in the beginning were law firms and professional service providers like PwC. We now have a lot of enterprise customers as well. And the way that the product is evolving is you can kind of think of it as actually two products. One is a productivity suite, right? And that is lawyers in the loop at all times, et cetera.

[16:33](https://realalpha.app/v/eXK-_yyQDMM/?t=993) And, you know, the ROI on that is incredibly helpful. It saves you tons of hours a week, et cetera, right? The other side of that is you're building these workflows that do part of the work from start to finish, right? And that is closer to selling the work. The way that we're actually approaching this is we are building those with law firms and helping them get more business, right? So we will get these revenue split agreements with law firms and professional services, and we will combine their domain expertise with our tech and then they go out and sell it to their clients, right?

[17:07](https://realalpha.app/v/eXK-_yyQDMM/?t=1027) And so we are transitioning from just a seat-based company to actually selling the work as well. And I think we'll see a lot of that this year. And what are the ripple effects of that, kind of like inside the building and outside the building? How do you manage that as a company and what change does that require of your customers? Yeah. So managing it internally is all about taking bets because it's, you know, if you're building software that is and every single feature that you're adding is good for your entire user base, you've kind of like reducing your chance of failure. Right. Because every time you add something to the software, it is hopefully increasing the value that you get from every single one of the users.

[17:52](https://realalpha.app/v/eXK-_yyQDMM/?t=1072) If you focus on building one of these specific workflows and it doesn't work, it's zero and one. It either works or it doesn't, right? And it's for a specific use case. The value of it is really high. But if you can't sell it to people, then you're in trouble, right? And so the best way that we've handled this internally is we've actually let the law firms do a lot of the discovery for us, right? So we will do joint projects with a large company and their law firm.

[18:23](https://realalpha.app/v/eXK-_yyQDMM/?t=1103) And so we will use them as design partners to make sure that this is something that is repeatable. You can actually do it. The technology is there to do it. And number three is appetite. So there is a large problem in legal where a lot of what you need is compliance checks or you're doing the work for insurance, et cetera. And so the third piece is. It's also really important where you have to make sure that the in-house team is actually okay with AI handling that use case, right?

[18:52](https://realalpha.app/v/eXK-_yyQDMM/?t=1132) So that's kind of how we handle it internally, where we spend tons of time on discovery, right? Tons of time talking to customers, and we have a lot of design partners for the early stages of that. One more piece about how we handle it internally as well. In order to generalize what we're doing is we're building, like, AI patterns. So here are the 15 types of actions in the legal industry that we need. Right. So research like case law research is a huge one. Regulatory research is another one. Clause extraction is another one. Right. And so we'll put tons of effort into those like kind of generic or widespread horizontal things. And then we'll chain them together kind of like a Ford factory line.

[19:31](https://realalpha.app/v/eXK-_yyQDMM/?t=1171) So those are the two ways. So on the GTM side, it's discovery. On the product side, it really is kind of building that Ford factory line of AI patterns. Externally, I think this will have a very large impact. So we're working with a firm that is almost 100 years old, and they became famous from literally advising, I think it was King Edward the something, on the abdication of the throne. So in other words, these firms are really old.

[19:59](https://realalpha.app/v/eXK-_yyQDMM/?t=1199) Like some of these firms are like hundreds of years old, and they have had the same business model for 100 years, right? And they're actually willing to work with us and change some of their business model, right? And like take a bet on this and explore it, right? And so as I think most people know, a lot of professional services, and especially legal, has been the billable hour forever, right? And so the problem with that is if you're selling them efficiency software, there's only so many hours in the day, right?

[20:26](https://realalpha.app/v/eXK-_yyQDMM/?t=1226) Unless you come up with a software that invents the 25th hour of the day, which, by the way, would be the best thing you could do to sell the professional services. But if you can't do that, efficiency is a hard sell unless you can convince them that there are ways for them to actually transform their business, right? So what we're starting with is let's help you take a bunch of these workflows that you normally do at not a very good cost or a loss and let's turn those into software and help you kind of spread out and get more market share in areas that you don't normally have.

[21:02](https://realalpha.app/v/eXK-_yyQDMM/?t=1262) So I'll give you a pretty good example of this. In private equity, one of the ways that law firms go and kind of get new private equity business is they'll do things like side letter compliance or kind of like lower end work. And they'll do that at a loss so that eventually the private equity firm will pay them for the LBO or whatever the big M&A deal is or something like that. And so we'll build them software to help them get that work and not operate at a loss to get it, charge for a flat fee, et cetera, and then take some of the cost.