Winston Weinberg, CEO of Harvey, a legal AI startup, has built one of Silicon Valley’s most talked-about companies in just a few years. Despite the niche nature of legal tech, Harvey has attracted top-tier investors including the OpenAI Startup Fund, Sequoia Capital, Kleiner Perkins, Andreessen Horowitz, Google Ventures, Coatue, and Elad Gil. The company’s valuation surged from $3 billion in February 2025 to $8 billion by late October, fueled by rapid adoption and strong financial performance. Weinberg, a first-year associate at O’Melveny & Myers, first saw the potential of AI in law while using GPT-3 to tackle a landlord-tenant case he knew nothing about. His co-founder Gabe Pereyra, then at Meta, helped develop a sophisticated chain-of-thought prompt based on California statutes. When tested on 100 real-world legal questions from Reddit, the AI-generated answers were deemed acceptable by two out of three attorneys without edits in 86 cases—proof that AI could deliver professional-grade legal work. That moment led to a pivotal cold email to Sam Altman and Jason Kwon at OpenAI on July 4th. They responded immediately, leading to a call with OpenAI’s leadership and a quick investment from the OpenAI Startup Fund. This early backing opened doors to angel investors like Sarah Guo and Elad Gil, setting Harvey on a fast growth trajectory. Weinberg, who had no background in tech or venture capital, credits the company’s success to focusing relentlessly on product and business performance. “The best way to raise money is to make sure your company is doing super well,” he said. “Spend 99% of your time on the business, and only a little on finding the right partners.” Harvey now has 235 clients across 63 countries, including most of the top 10 U.S. law firms, and has surpassed $100 million in annual recurring revenue. The company is expanding rapidly, with corporate clients now making up 33% of revenue—up from 4% at the start of the year. A key driver has been law firms themselves, who are introducing Harvey to their corporate clients, creating a network effect. A major focus for Harvey is its “multiplayer” platform, which enables secure collaboration between in-house legal teams and outside law firms. This requires solving complex issues like data residency, ethical walls, and permissioning across multiple jurisdictions. Germany and Australia, for example, have strict data laws requiring financial data to stay within borders. Harvey has built infrastructure in each country, though early costs were high due to underutilized compute capacity. The company’s primary use cases are drafting, legal research (enhanced by a LexisNexis partnership), and document analysis—especially in due diligence and discovery. While transactional work like M&A remains strong, litigation is growing fast, driven by access to large volumes of case data. Critics say Harvey is just a ChatGPT wrapper, but Weinberg argues the real moat lies in two areas: workflow data and the multiplayer platform. By collecting how legal AI is used in real-world scenarios, Harvey is building evaluation systems that improve accuracy over time. And unlike competitors focused only on law firms or only on in-house teams, Harvey is building a unified system that connects both sides. The business model is currently seat-based, but the company is moving toward outcome-based pricing for high-accuracy, high-value tasks. However, Weinberg stresses that AI will not replace lawyers—especially not junior ones. Instead, he sees AI as a powerful tutor, offering real-time feedback and guidance, turning legal training into a more efficient, scalable process. Despite its rapid growth, Harvey has no immediate plans for another large fundraising round. The company is cash-flow positive and not burning through capital. While public markets are a long-term goal, Weinberg says there’s no timeline yet. At its core, Harvey is still in early innings. With only a tiny fraction of the world’s 8 million lawyers using the platform, and legal work still far from fully automated, the potential for AI in law remains vast. As Weinberg puts it, the value per token in a high-stakes merger agreement could be millions of dollars—making the future of legal AI not just promising, but transformative.