On this page
Harvey is the most prominent generative AI platform built specifically for legal work. Founded in 2022 by Winston Weinberg, a former O’Melveny & Myers litigator, and Gabriel Pereyra, a research scientist who worked at Google Brain and DeepMind, it was the OpenAI Startup Fund’s first major legal bet. Allen & Overy (now A&O Shearman) rolled it out firm-wide in 2023, PwC signed a landmark deal, and successive funding rounds pushed Harvey’s valuation into the billions.
The product is more than a chatbot with a law degree. Assistant handles research, drafting, and summarization with legal-domain tuning built on OpenAI models. Vault lets teams upload thousands of documents — a data room, a contract portfolio — and run structured queries across all of them, which is where due diligence hours actually disappear. Workflows encode repeatable multi-step tasks so associates get consistent output rather than freeform prompting, and a Microsoft Word add-in brings drafting help into the document itself.
Harvey’s main rivals are Thomson Reuters’ CoCounsel (built from the Casetext acquisition), Sweden’s Legora, Robin AI, and Spellbook at the small-firm end. Its differentiators are deep firm partnerships and security posture — client data stays siloed and isn’t used to train shared models.
Key Features
- Assistant for legal research, memo drafting, and document summarization
- Vault — query and extract across thousands of uploaded documents at once
- Workflows for standardized multi-step tasks like diligence checklists
- Contract review with redline suggestions and clause comparison
- Microsoft Word add-in for in-document drafting assistance
- Enterprise security: SOC 2, data isolation, no training on client data
- Jurisdiction-aware research spanning case law, statutes, and regulations
Pricing
- Enterprise: Custom, quote-based per-seat contracts negotiated with firms and legal departments. No free tier, no self-serve signup, no published rates — reported figures suggest a premium price per lawyer per year.
Best For
Big Law firms, in-house teams at large companies, and professional services firms doing document-heavy work — M&A due diligence, portfolio contract review, regulatory surveys — where shaving associate hours at scale justifies enterprise pricing.
Limitations
Harvey is closed to the lawyers who might benefit most: solos and small firms can’t buy it, and Spellbook or CoCounsel serve that market instead. And like every legal AI, its output is a first draft — hallucinated citations remain a career-level risk, so everything still passes through attorney review.
