\n
\n\t\t\t\t\t\t\t\t\tSan Francisco<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\t|<\/span>
\n\t\t\t\t\t\t\t\t\t\t\t\t\tOctober 13-15, 2026<\/span>\n\t\t\t\t\t\t\t<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\nWhat happened next?<\/strong><\/p>\nWe cold emailed Sam Altman and Jason Kwon, who was the general counsel at OpenAI. We figured we had to email a lawyer because otherwise the person wouldn\u2019t know if the outputs were right. On the morning of July 4th at 10am \u2014 I remember this specifically because it was July 4th \u2014 we got on a call with them and kind of the rest of the C-suite at OpenAI, and we made our pitch.<\/p>\n
Did they write a check right away?<\/strong><\/p>\nYeah. It\u2019s the OpenAI Startup Fund [they are the second-largest investor in Harvey]. OpenAI introduced us to our angel investors at the time, Sarah Guo and Elad Gil, and then everything else from there we were doing ourselves. I actually didn\u2019t have any friends that worked in tech. I didn\u2019t grow up in San Francisco. I didn\u2019t know who the top VCs were. I didn\u2019t understand how you\u2019re supposed to fundraise. This was all just net new to me.<\/p>\n
For someone who wasn\u2019t familiar with the VC scene, you\u2019ve raised a lot of money. What enabled you to raise so much?<\/strong><\/p>\nI might say something the VC community might not love, but I strongly believe that the best way to raise money is to just make sure your company is doing super well. I think there\u2019s a lot of advice out there about networking, but to me, the most important thing is to spend almost the entire time on your business and then find VCs that want to do that with you. You need to find a few partners that you think are going to go the distance with you. 99% of your time, focus on the business going well, and then spend time trying to find a few folks that you really think you can partner with and that will be there for you for the long run.<\/p>\n
You hit $100 million in ARR in August. With around 400 employees, how close are you to break-even?<\/strong><\/p>\nCompute costs are more expensive for us than a lot of other things. We\u2019re operating in more than 60 countries with data residency laws in all of them. For a long time, if you used multiple models in your product, you had to buy a bucket of compute \u2014 a minimum threshold \u2014 in every single one of those countries, even if you didn\u2019t have enough clients yet to support that cost.<\/p>\n
Germany and Australia have incredibly strict data processing laws. You cannot send financial data outside of those countries. We\u2019d set up Azure or AWS instances in every single one of those countries, but we\u2019d only use them to close three or four large clients. Our margins look very good on a token basis, but they\u2019re worse because we have to spend so much on upfront compute across so many jurisdictions. That will get solved over time.<\/p>\n
Tell us about your sales process. How are you expanding globally?<\/strong><\/p>\nAt the beginning of this year, about 4% of our revenue was from corporates and 96% from law firms. Right now, 33% of our revenue is from corporates, and my gut is by the end of the year, that looks closer to 40%.
In the beginning, we would take public litigation briefs from Pacer, find the partner that wrote it, put them into Harvey, and show them how they could argue against their own brief. That got massive attention because it was relevant to what they just did. <\/p>\n
But what was interesting is once we got adoption at law firms, the law firms themselves would help us pitch to corporates. A firm like Latham will introduce Harvey to clients and say, \u201cHey, did you know this is how we can use AI to do XYZ?\u201d So what started happening was law firms would actually help us sell to corporates because they want to collaborate in the system.<\/p>\n
You refer to this as \u201cmultiplayer.\u201d Can you expound on this a a growing area of focus?<\/strong><\/p>\nThis is a huge problem. You\u2019ve seen announcements from OpenAI and Microsoft about shared threads and company memory. That\u2019s hard \u2014 you have to get the permissioning right so agents can access the right systems. But you\u2019re only solving it for one entity at a time.<\/p>\n
The secondary problem we have is: how do you solve that for a company plus all its law firms? You need to get the permissioning right internally and externally. There\u2019s a concept in legal called ethical walls. Think about a law firm in the valley that works with 20 VCs. If you\u2019re working on a deal for Sequoia but also working on another deal for Kleiner Perkins, what happens if you accidentally give all the data on the Sequoia deal to Kleiner Perkins? Huge, astronomical problem. We have to solve internal permissioning and external permissioning so agents can work correctly, and if you get it wrong, you\u2019re going to have disastrous impacts on the industry.<\/p>\n
Have you solved this?<\/strong><\/p>\nIt\u2019s definitely in process. We\u2019re doing all of the security and the permissioning first. The first version of this at scale will probably be done in December. The nice thing is because such a high percentage of our customer base are already corporates using Harvey, the security problem is much easier because they\u2019ve already gone through security review.<\/p>\n
How are<\/em> lawyers primarily using Harvey today?<\/strong><\/p>\nIn this order: Number one is drafting. Number two is research \u2014 that\u2019s emerging because we just have a partnership with LexisNexis. And the third is analyze. What I mean by analyze is running 10 questions over 100,000 documents, like what you do in diligence or discovery.<\/p>\n
In the beginning, we had much more transactional use cases \u2014 M&A and fund formation. Those are still very popular, and we\u2019re building modules specifically for those matters. The area that\u2019s growing faster is litigation, and a lot of that is because you needed the data before you could do it.<\/p>\n
Some critics have said Harvey is just a wrapper for ChatGPT. How do you respond?<\/strong><\/p>\nThe largest advantage we have over time is two things. One, we\u2019re collecting a tremendous amount of workflow data \u2014 what are the main use cases these models can actually do? Evaluation becomes a pretty strong moat because how do you evaluate the quality of a merger agreement? That becomes really hard. You have to set up evaluation frameworks and agentic systems that can self-eval all the different steps.<\/p>\n
The second strongest moat is our product is becoming very strongly multiplayer. This industry has two sides \u2014 providers of legal services and consumers. You need to build a platform that\u2019s in between both. So far, I haven\u2019t seen a competitor doing that. We have competitors doing what we do for law firms, and competitors doing what we do for in-house, but I haven\u2019t seen someone build a truly multiplayer platform.<\/p>\n
In terms of the \u201cChatGPT wrapper\u201d criticism \u2014 for 2023 and 2024, a lot of the power behind the product is honestly the model plus front-end work that makes the UI and UX easier. But if you\u2019re trying to build something where I have 100,000 documents in this data room, 5,000 emails about this M&A, all these different statutes and codes, and I want a system where I can ask questions over all of those pieces combined with high accuracy \u2014 that\u2019s the holy grail. We\u2019ve created all the pieces, and what we\u2019ve been building for the past couple months is pulling that together. <\/p>\n
What\u2019s your business model?<\/strong><\/p>\nRight now it\u2019s mostly seats, but we\u2019re moving to more outcome-based pricing as the workflows get more complex. You want to do both. You want outcome-based pricing for very small things that you can ensure have the exact same level of accuracy as a human, or better, with very high speed. But the reality is you\u2019re going to want a lawyer in the loop for so much of work.<\/p>\n
For at least the next year or two, it\u2019s a productivity suite sold seat-based and multiplayer between law firms and their in-house teams. Slowly over time, we\u2019ll build more consumption-based workflows as the systems get better and more accurate than humans in some areas. But it\u2019s not going to be like you automate an entire M&A \u2014 it\u2019s going to be specific pieces of diligence where you can have disclosure agents automate the first pass, then have lawyers jump in and do the rest.<\/p>\n
You mentioned to us earlier that penetration is really low in legal. How low?<\/strong><\/p>\nWhat percentage of the lawyers on Earth are using Harvey right now? It\u2019s a super low percentage. There are 8 or 9 million lawyers on Earth. But the more interesting point is we are unbelievably early innings on how complex of work these systems can do. They\u2019re very helpful and people are getting incredible ROI, but if you think about what percentage of legal work can these systems do today versus what I think it can do in the next five years \u2014 it\u2019s so much lower.<\/p>\n
Think about the use case as, what is the value per token. The legal fees for a merger could easily be tens of millions of dollars. The artifact you have after that merger is a merger agreement and an SPA \u2014 maybe 200 pages total. What is the value per token on that document that required $20 million or $30 million of legal fees to generate? Those are the types of use cases where, when I say we\u2019re at incredibly low penetration, it\u2019s that we aren\u2019t at the point where you can do something like that. And the value of being able to do that accurately is incredibly high.<\/p>\n
What happens to junior lawyers who are no longer getting the apprenticeship they might have had in the past?<\/strong><\/p>\nI care about this potentially more than anything else at the company because I was a junior lawyer very recently. The goal of law firms in the next five to ten years is: how fast can you train the best partners? I think right now, that\u2019s partially the goal, but partially the goal is we hire armies of associates and bill them out a lot. Whether it\u2019s because things become outcome-based pricing or because partners can charge more if AI systems can\u2019t do what they do, the most important thing financially for a law firm is to make sure you\u2019re hiring, training, and developing lawyers that get to being a partner as fast as humanly possible.<\/p>\n
If you can build tools that can do the first pass of an M&A, that is a one-on-one tutor for a junior associate. We work with a lot of law schools. You can imagine at some point you have an AI merger that you do in Harvey \u2014 the system\u2019s teaching you, giving you real-time feedback. That\u2019s an incredible training system. If you can build systems that can actually do a lot of the tasks, there\u2019s no reason you couldn\u2019t turn that into one of the best education platforms possible.<\/p>\n
With your valuation jumping from $3 billion to $8 billion in less than a year, what are your plans for future fundraising?<\/strong><\/p>\nFundraising large rounds is not something we have planned anytime soon. We don\u2019t need that much money, and we aren\u2019t burning a crazy amount. The reason I did a lot of fundraising this year is there are research directions that are going to require a lot of compute, and we wanted to prepare ourselves for that. In terms of public markets, that\u2019s definitely what we\u2019re interested in long term. I can\u2019t give you anything close to a timeline, but we\u2019re interested.<\/p>\n<\/div>\n
\n
Source link <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"Legal AI might not sound like the sexiest category in Silicon Valley, yet Harvey\u2018s CEO Winston Weinberg has captured the attention of virtually every top-tier investor in the Valley. The company\u2019s backers read like a who\u2019s who of venture capital: the OpenAI Startup Fund (its first institutional investor), Sequoia Capital, Kleiner Perkins, Elad Gil, Google […]<\/p>\n","protected":false},"author":1,"featured_media":205179,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[14],"tags":[],"class_list":{"0":"post-205178","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-tech"},"_links":{"self":[{"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/posts\/205178","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/comments?post=205178"}],"version-history":[{"count":0,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/posts\/205178\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/media\/205179"}],"wp:attachment":[{"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/media?parent=205178"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/categories?post=205178"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/entertainment.runfyers.com\/index.php\/wp-json\/wp\/v2\/tags?post=205178"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}