In a regulated market where accuracy is non-negotiable, pair a domain expert with a technical one, win one marquee anchor first, and let its prestige sell everyone else.
Harvey
Harvey is a domain-specific AI platform for lawyers, founded in the summer of 2022 by litigator Winston Weinberg and AI researcher Gabriel Pereyra. Its rise is a case study in choosing a hard, regulated market on purpose: rather than ship a horizontal chatbot, Harvey picked elite law as its entry point, paired legal and machine-learning expertise at the founding table, and let the most prestigious firms vouch for it. The three stories below trace that single strategic choice from wedge, to team, to sales motion.
Beachhead: winning elite BigLaw before going broad
The problem. In late 2022, general-purpose models like ChatGPT were exciting but untrustworthy for legal work: they hallucinated citations and gave regulated, risk-averse buyers no reason to rely on them. A horizontal "AI for everyone" wedge would have put Harvey in a crowded fight with no credibility in the one profession where accuracy is non-negotiable.
The approach. Harvey narrowed to a single high-value regulated vertical, elite corporate law, and landed a marquee anchor: it signed Allen & Overy (one of the UK Magic Circle firms, now A&O Shearman) as an exclusive launch partner, having quietly trialed the product inside A&O's Markets Innovation Group since November 2022.
How it solved it. In February 2023 A&O rolled Harvey out to more than 3,500 lawyers across 43 offices, after a trial in which those lawyers had already run roughly 40,000 queries for day-to-day client work. It was reported as the first known use of a generative AI product inside a Magic Circle firm, giving Harvey a defensible foothold and instant credibility with exactly the buyers it wanted next.
Founder-Market Fit: a litigator and a DeepMind researcher as roommates
The problem. Building legal AI that lawyers will actually trust demands two rare things at once: deep, practitioner-level understanding of how legal work is really done, and the ability to build and fine-tune frontier models. Most founders have one or the other, and a team missing either half tends to ship either a toy or an untrustworthy tool.
The approach. Harvey was co-founded by exactly that pairing. Winston Weinberg was a securities and antitrust litigation associate at O'Melveny & Myers, and Gabriel Pereyra was a research scientist who had worked on AI at Google DeepMind and Meta. The two were roommates in San Francisco when they started the company in 2022.
How it solved it. The combination convinced the OpenAI Startup Fund to make Harvey one of its first investments, an approximately $5 million seed round in late 2022. That credibility and capital let Harvey build models tuned to legal workflows rather than a generic chat wrapper, the very thing that made A&O and later firms comfortable deploying it.
GTM: land the top of the market, then let it refer
The problem. Legal buyers are conservative, procurement is slow, and a new vendor has no track record to point to. Harvey needed a sales motion that could scale into a skeptical industry without burning years cold-calling every firm.
The approach. Harvey ran a top-down land-and-expand play: win the most prestigious firms first so their names became the sales pitch, then grow through their referrals rather than outbound alone. Reportedly, its first 50 enterprise customers all came as referrals from existing law-firm clients, and it deliberately targeted the top of the market (a majority of the top 10 US firms) to bank credibility.
How it solved it. The motion compounded fast: Harvey grew from roughly 40 customers in early 2024 to about 235 across 42 countries by year-end, reaching an estimated $65.8 million in revenue in 2024 (a 558% year-over-year jump) and crossing the $100 million ARR mark by August 2025, while expanding from BigLaw into corporate legal departments and professional-services buyers like PwC.