Harvey is an enterprise AI platform for legal work, used by 100,000+ lawyers at 1,300+ organizations to draft, review, and analyze contracts, briefs, and legal documents using domain-specific AI agents.
A reasoned read from public sources. Each point links to its source.
Harvey is the clearest example of vertical AI done right: they picked the hardest, most compliance-sensitive buyers, earned trust through domain expertise rather than demos, and built switching costs through workflow encoding rather than just features. The founding team's complementary backgrounds — a practicing litigator and a DeepMind researcher — gave them genuine insight into why generic AI failed in legal, not just a market opportunity. The risk is not whether Harvey is a real company (it clearly is), but whether the $11B valuation leaves enough upside for late-stage employees and whether the agent platform bet pays off before a well-resourced incumbent (Thomson Reuters, Microsoft, or OpenAI itself) closes the gap.12345
The U.S. legal market bills $150B+ annually, and Harvey is targeting the highest-value segment — BigLaw and in-house enterprise legal teams — where AI-driven efficiency gains are most monetizable. Sacra estimates Harvey's TAM expansion includes 90% of legal professionals currently unreachable at current pricing, suggesting the addressable market is both large and largely untapped. The company operates in 60 countries with 1,400+ customers, and the legal AI category is growing rapidly as frontier models improve. However, the pricing floor structurally locks out the SMB legal market, concentrating revenue risk in a relatively small number of large enterprise accounts.56789
Harvey's product has evolved from a chat-style legal assistant to a full agentic platform with 25,000+ custom workflow agents deployed across client environments — each one a compounding switching cost. The Vault architecture solves the core enterprise objections (data isolation, attorney-client privilege, hallucination) that made generic AI 'legally radioactive,' with a reported 0.2% error rate via citation-backed generation. The LexisNexis content deal closed a critical data credibility gap, and the Agent Builder enables clients to encode their own workflows, making displacement progressively harder. The strategic shift from co-pilot to agent platform is the right direction, but multi-step agent coordination across chained LLMs introduces meaningful reliability risk at scale.10371112
Co-founder Winston Weinberg is a former securities and antitrust litigator who left BigLaw after one year — giving him authentic domain credibility with the exact buyers Harvey targets. Co-founder Gabriel Pereyra brings serious AI research pedigree: Google Brain, DeepMind, and Meta AI, with a CS degree from USC. The origin story — Pereyra showing GPT-3 to his lawyer roommate Weinberg — is a genuine insight-driven founding, not a market-map exercise. The GTM motion of hiring 'Legal Engineer AEs' (practicing attorneys from Vault 50 firms) to close deals through peer credibility rather than sales persuasion is a sophisticated and defensible go-to-market insight. Headcount grew 140% YoY to 781 people, suggesting rapid organizational scaling that carries its own execution risk.1132118
Harvey's revenue trajectory is exceptional by any measure: $0 to $195M ARR in approximately three years, with 290% YoY growth from $50M in 2024. The company crossed $100M ARR in August 2025, with weekly active users quadrupling YoY and monthly queries growing 5.5x. The customer base spans 1,400+ organizations across 60 countries, including 45+ AmLaw 100 firms and 50 asset managers, with over 100,000 lawyers actively using the platform. Estimated NRR above 130% suggests strong expansion within accounts. The $1.226B raised across 11 rounds — with Sequoia backing Harvey three times — is a strong signal of institutional conviction.1415164911
The competitive landscape includes Legora (collaborative legal AI, strong in Europe), Hebbia (finance and legal workflows), Ivo (contract review), Spellbook, and a long tail of point solutions. Thomson Reuters and LexisNexis — the $5B/year incumbents — are the most dangerous long-term threats given their data moats and existing firm relationships. Harvey's strategic response has been to partner with LexisNexis on content rather than fight them directly, while embedding engineering teams inside accounts to compress churn. The pricing floor that locks out 90% of the legal market is simultaneously Harvey's biggest competitive vulnerability (leaving room for lower-cost entrants) and its clearest signal that it is deliberately choosing enterprise defensibility over market breadth.17187919
Harvey has built the most defensible position in legal AI by targeting the hardest buyers first (Magic Circle and AmLaw 100 firms), earning trust through domain-expert sales, and then encoding client workflows into 25,000+ custom agents that create compounding lock-in. The 290% YoY ARR growth, >130% estimated NRR, and Sequoia's three-time backing collectively suggest this is not a hype cycle — it is genuine enterprise adoption. The agentic platform shift, if it works, transforms Harvey from a productivity tool into the operating system for legal work, dramatically expanding per-seat economics and making displacement nearly impossible for entrenched accounts.911476
At an $11B valuation and ~57.9x EV/Revenue, Harvey must sustain hypergrowth for years to justify the multiple — and the growth rate will mathematically compress as the base grows. The pricing floor structurally excludes 90% of the legal market, concentrating revenue in a small number of large accounts where a handful of churn events could be catastrophic. Harvey scrapped its fine-tuned legal model when frontier reasoning models commodified legal reasoning, revealing a core dependency on OpenAI that could become a liability if the relationship sours or if OpenAI moves further into vertical applications. Multi-step agent coordination across chained LLMs introduces reliability risk that, in a zero-tolerance legal environment, could trigger high-profile failures and reputational damage.2079125
Harvey needs to successfully transition from co-pilot to agent platform — proving that multi-step legal agents can handle complex, high-stakes workflows like fund formation and international M&A without reliability failures that damage firm reputations. It must expand beyond BigLaw into mid-market legal and in-house teams (currently the 90% of the market locked out by pricing) without diluting the enterprise trust moat. The OpenAI relationship must remain strategic rather than competitive, and Harvey must build enough proprietary workflow data and client-specific customization that model commoditization strengthens rather than weakens its position. Finally, it needs to sustain >100% NRR as it scales past $200M ARR to justify the valuation multiple.127911
If you're a senior engineer at big tech, Harvey is one of the most credible vertical AI companies to consider — but the calculus depends heavily on your stage expectations. At $11B and $195M ARR, this is late-stage growth equity territory, not early startup. The equity upside requires Harvey to reach $3-5B+ ARR and sustain a premium multiple, which is achievable but not guaranteed. The technical problems are genuinely hard: multi-step legal agent reliability, retrieval over massive document corpora, and building infrastructure that works in a zero-hallucination-tolerance environment. If you want to work on consequential AI infrastructure with real enterprise customers and a domain-expert team, Harvey is compelling. If you're optimizing for 10-100x equity returns, the risk/reward at this valuation is more like a pre-IPO bet than a startup swing.168411320
Harvey is an enterprise AI platform for legal work, used by 100,000+ lawyers at 1,300+ organizations to draft, review, and analyze contracts, briefs, and legal documents using domain-specific AI agents.
Founded in 2022 by Winston Weinberg, a former O'Melveny securities litigator, and Gabriel Pereyra, a former research scientist at Google DeepMind and Meta. After experimenting with GPT-3, the pair cold-emailed Sam Altman and Jason Kwon at OpenAI in 2022, leading to an early check from the OpenAI Startup Fund.
Fortune profile of 30-year-old CEO behind $11B legal AI startup.
A16z-led round confirms $11B valuation; Harvey now at $190M ARR.
ARR grew from $100M (Aug) to $190M (Jan), with 100K+ lawyers across 1,300 organizations.
Growth round co-led by GIC and Sequoia values Harvey at $8B.
TechCrunch deep-dive into Harvey's origin and culture.
Series E co-led by Kleiner Perkins and Coatue.
Series D at $3B valuation to scale legal AI agents.
Co-founder & CEO
Previously Securities Litigation Associate at O'Melveny & Myers
Co-founder and CEO of Harvey. Former first-year securities litigation associate at O'Melveny who turned a side project with GPT-3 into a $11B legal AI startup.
Co-founder & President
Previously Research Scientist at Google DeepMind
Co-founder and President of Harvey. Previously research scientist at Google DeepMind and Meta AI.
$1.2B raised total
Petitioner on record
HARVEY NASH DBA LATITUDE 36 INC · FRANKLIN, TN
FY 2025
See full record