The leading enterprise legal AI platform, used by 80% of the Am Law 100 and five of the Fortune 10. It has evolved from a legal chat assistant into an agentic platform: a library of 500+ lawyer-built agents covering M&A, funds, litigation, IP, and investigations, each able to produce Word/PPT/Excel deliverables; an Agent Builder for customizing agents in plain English; Harvey II Memory that learns each lawyer’s drafting style and citation preferences across Harvey, Word, and Outlook; and Harvey Graph / Fact Graph (Nov 2026) that connects facts across millions of documents for disputes and investigations, linking back to source text and flagging inconsistencies. An Outlook add-in handles email triage, drafting, and review initiation, with human approval required for every action. Pricing: Quote-only, enterprise sales, no free tier or self-serve trial. Third-party reports suggest ~$1,000–$2,000/seat/month at mid-market with 25–50 seat minimums, dropping at Am Law 100 scale. SOC 2 Type II, ISO 27001/27701/42001 certified. Raised $550M at a $15.5B valuation (Sept 2026).

Pros & Cons:

Pros: The domain-specific agent library is unusually deep—500+ agents built and pressure-tested by Harvey’s internal lawyers, covering practice areas from M&A to white-collar investigations, with outputs in Word/PPT/Excel rather than just chat text; Harvey II Memory learns individual lawyer preferences and carries them across Word and Outlook, which addresses the “start from scratch every time” problem that plagues generic AI tools; the Fact Graph for litigation (Nov 2026) connects facts across millions of documents and flags inconsistencies as new evidence arrives, a capability that goes well beyond document search; named enterprise customers (80% of Am Law 100, Microsoft’s legal department) and SOC 2/ISO 42001 certifications provide strong procurement validation.

Cons: No public pricing, no free tier, no self-serve trial—every deployment requires a sales conversation and custom quote, making cost comparison impossible before a demo, and reported seat minimums (25–50) exclude small firms entirely; reported costs are very high for mid-market—~$1,000–$2,000/seat/month with annual commitments and potential 10–25% renewal uplifts, which is prohibitive for firms without enterprise budgets; the product surface is now very broad (agents, Memory, Fact Graph, Outlook, iManage, Graph), and understanding which modules are included vs. separately priced requires careful scoping during sales; it is a closed enterprise platform—no self-hosted option, no public API for external developers (MCP connectors are early access and governed by firm admins); independent third-party validation of specific agent accuracy is limited—most evidence comes from Harvey’s own materials and named customer testimonials, not independent benchmarks.

Who is it for?:Large law firms (Am Law 100 scale) and enterprise in-house legal departments that need a governed, enterprise-grade legal AI platform across multiple practice areas; litigation and investigations teams that need to connect facts across millions of documents (Fact Graph); firms already using iManage that want two-way sync with ethical walls enforced; organizations that require SOC 2, ISO 27001, and ISO 42001 certification for legal AI procurement. Not suited for solo practitioners or small firms (25–50 seat minimums reported, no public pricing), teams seeking a self-serve trial or free tier, or anyone needing a simple contract-drafting tool at a published price (Spellbook is a better fit).