Harvey review
Domain-tuned AI for law firms: research, drafting, diligence, and agentic legal work.
TECHi verdict
Harvey won the credibility race in legal AI — OpenAI backing, an $11 billion valuation, and 100,000-plus lawyers are not accidents — but buyers should be clear-eyed about what the premium buys. Reported pricing of $1,200-plus per seat per month is several times a general-purpose enterprise AI seat, and the justification is workflow fit and confidentiality posture, not raw model quality; the underlying intelligence is largely OpenAI's. For a large firm running diligence at volume, the math can genuinely work — one saved associate-week per matter covers a lot of seats. For small firms and solos, CoCounsel or Spellbook delivers most of the drafting value at a fraction of the cost. And no vendor, Harvey included, has made unsupervised legal output defensible: a lawyer still signs everything.
Pros
- + Purpose-built legal workflows — agentic diligence, matter workspaces, firm-specific tuning — that generic assistants don't replicate
- + Adoption across 1,300+ organizations including, per industry reporting, about half the Am Law 100 — reference customers rivals can't match
- + Confidentiality architecture and security posture designed for privileged material, the threshold requirement for legal buyers
Watchouts
- - Reported $1,200+/seat/month with ~20-seat minimums and no published pricing prices out small firms entirely
- - Citation accuracy remains legal AI's open wound — every output needs qualified attorney review before it touches a filing
Frequently asked
What are the pros of Harvey?
Purpose-built legal workflows — agentic diligence, matter workspaces, firm-specific tuning — that generic assistants don't replicate. Adoption across 1,300+ organizations including, per industry reporting, about half the Am Law 100 — reference customers rivals can't match. Confidentiality architecture and security posture designed for privileged material, the threshold requirement for legal buyers
What are the watchouts / cons of Harvey?
Reported $1,200+/seat/month with ~20-seat minimums and no published pricing prices out small firms entirely. Citation accuracy remains legal AI's open wound — every output needs qualified attorney review before it touches a filing
What is the TECHi verdict on Harvey?
Harvey won the credibility race in legal AI — OpenAI backing, an $11 billion valuation, and 100,000-plus lawyers are not accidents — but buyers should be clear-eyed about what the premium buys. Reported pricing of $1,200-plus per seat per month is several times a general-purpose enterprise AI seat, and the justification is workflow fit and confidentiality posture, not raw model quality; the underlying intelligence is largely OpenAI's. For a large firm running diligence at volume, the math can genuinely work — one saved associate-week per matter covers a lot of seats. For small firms and solos, CoCounsel or Spellbook delivers most of the drafting value at a fraction of the cost. And no vendor, Harvey included, has made unsupervised legal output defensible: a lawyer still signs everything.