Last updated: 2026-09-03
Harvey is an enterprise legal AI platform from Harvey (Counsel AI), founded in 2022 and used, by its own count, by more than 2,400 legal organisations, 200,000 professionals and 75 of the AmLaw 100 across 70 countries. Its products are Agents, Vault for bulk document work, Knowledge, Spaces, Command Center and Contract Intelligence, built on Azure OpenAI models fine-tuned on legal domain data, with answers cited to source documents. It is sold enterprise-only through a demo request, with no public price list. Harvey contractually guarantees it does not train models on customer inputs, outputs or uploaded documents, and publishes SOC 2 Type II, ISO 27001, ISO 27701 and ISO 42001.
About Harvey
Harvey is a generative AI platform purpose-built for law firms and in-house legal teams. Winston Weinberg, a former O'Melveny & Myers litigator, and Gabriel Pereyra, a former Google DeepMind researcher, founded the company in summer 2022. Harvey has raised $1.22 billion across ten rounds and was valued at $11 billion in its March 2026 Series G, led by GIC and Sequoia. The platform now serves 142,000+ legal professionals across 60 of the AmLaw 100 and Fortune 500 companies. Four products cover the legal lifecycle. Assistant handles document analysis and drafting with real-time legal research. Vault bulk-analyzes thousands of documents at once for due diligence and M&A. Knowledge answers complex regulatory and tax questions with citations tied to case filings and statutes. Workflow Agents execute multi-step legal work end to end, from contract review to risk scoring. Across all four products, Harvey processes 1.3 million documents and 200,000+ queries daily. Core use cases are contract review, due diligence document discovery and risk assessment, litigation document clustering and privilege review, and automated regulatory compliance checks, with review running roughly 80x faster than doing it by hand. The platform fits in-house counsel and large firms handling high-volume, high-stakes legal work; solo practitioners and small firms are priced out of its enterprise-only sales model entirely. Compliance runs deep: SOC 2 Type II (2026 audit), ISO 27001, ISO 42001 for AI management, GDPR, CCPA, and EU-US Data Privacy Framework certification, plus a zero-training guarantee on customer data and regional data residency in the EU, US, or Australia. Customers report 92% monthly active usage and save 20+ hours a month on average, evidence the tool earns its keep beyond the initial sales pitch.
Screenshots
Pricing
There is no public price list and no self-serve signup: Harvey sells enterprise contracts through a demo request, and the site publishes no figure. Buyer-reported ranges put a deployment in the tens to low hundreds of thousands of dollars a year depending on seat count, with minimum seat counts and an implementation phase on top, but those numbers come from customers rather than from the vendor and should be treated as hearsay until you have a quote. Vendor pages re-checked 3 September 2026.
Feature Comparison by Tier
| Feature | Enterprise |
|---|---|
| Price | Custom, quote only |
| Self-serve signup | — |
| Agents (multi-step legal work) | ✓ |
| Vault (bulk document review) | ✓ |
| Contract Intelligence | ✓ |
| Knowledge, Spaces, Command Center | ✓ |
| Citations to source documents | ✓ |
| Office integration (Word, Outlook, SharePoint) | ✓ |
| Mobile apps | ✓ |
| Training on customer data | Contractually excluded |
| Retention window | Customer-configurable |
| Certifications | SOC 2 II · ISO 27001 · 27701 · 42001 |
| Enterprise controls | SAML SSO, audit logs, IP allow-listing |
Key Features
- Contract Analysis at Scale: Processes 1.3 million documents daily with automated risk scoring, obligation extraction, and redline suggestions across every contract under review.
- Citation-Backed Legal Research: Every legal research answer traces back to the specific statute, case, or regulation it relies on, with sentence-level citations lawyers can click through to confirm before relying on the output.
- Workflow Agents for Complex Work: Pre-built and custom agents run multi-step legal work end to end, handling contract negotiation loops, due diligence workflows, and compliance reviews without a lawyer driving each step.
- Mobile Voice Dictation: iOS and Android apps, launched in September 2025, let lawyers dictate hands-free and scan documents from a phone, syncing query history across devices so work started on mobile continues at a desk.
- Vault for Bulk Document Analysis: Securely upload and organize legal documents, then analyze up to 10,000 documents per vault at once for discovery, M&A due diligence, and litigation support.
- MCP Integration Standard: Embeds Harvey's legal Q&A and Vault analysis directly into external AI agents through the Model Context Protocol, so tools like Claude can call Harvey without a separate login.
Pros
- The speed and volume Harvey handles turns document review from a bottleneck into a background task, freeing lawyers to spend billable hours on judgment calls instead of first-pass reading.
- Sentence-level citations mean lawyers can actually verify Harvey's research instead of treating it as a black box, which matters enormously for malpractice risk in legal work.
- Deep adoption inside large law firms gives Harvey a credibility and network-effect moat that newer legal AI entrants cannot easily match.
- The compliance certification stack (SOC 2, ISO 27001, ISO 42001) and a zero-training guarantee make Harvey easier to clear through a law firm's own security review than most legal AI vendors.
- Mobile and MCP support mean Harvey extends beyond the desktop, both for lawyers working out of office and for other AI agents that need to call into Harvey's legal analysis.
Cons
- Enterprise pricing puts Harvey out of reach for solo practitioners and small firms; implementation and training fees add significant overhead on top of the negotiated base license.
- The learning curve is steep, and Vault's hard document cap can force teams to split a single large litigation matter across multiple vaults; power users also report small glitches and a weak dashboard.
- Citations can still be wrong if the underlying model fails, so lawyers must keep fact-checking rather than treating any answer as final; this is a legal-judgment tool, not a fire-and-forget one.
- There's no self-serve path at all: no free trial, no public pricing, and every deal goes through enterprise sales, which locks out firms below a certain size before they can even test it.
Product Information
- Cloud
- Yes
- Self-Hosted
- No
- On-Premise
- No
- Languages
- English, Spanish, French, German, Chinese
- Training
- Implementation onboarding, Product documentation, Customer success webinars, Video tutorials, Professional services for custom agents
Data Handling
- Trains on your data
- No
- Training-data policy
- No. Harvey's security page states it contractually guarantees through its Platform Agreement that customer data stays the customer's, and that inputs, outputs and uploaded documents are not used to train underlying models. Retention is configurable by the customer rather than a fixed window, and the security page lists SAML SSO, audit logs, IP allow-listing and data lifecycle management as enterprise controls. Certifications shown on the site: SOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, GDPR and CCPA. Checked 3 September 2026.
- Data residency
- Customer-controlled. The security page frames data sovereignty as a customer decision over what is uploaded and where it lives; regional hosting is scoped in the enterprise contract rather than published.
- Compliance
- SOC 2 Type II · ISO 27001 · ISO 27701 · ISO 42001 · GDPR · CCPA
Frequently Asked Questions
What does Harvey actually cost?
Harvey publishes no price and has no self-serve signup: every deployment is a negotiated enterprise contract that starts with a demo request. Buyers report figures in the tens to low hundreds of thousands of dollars a year depending on seat count, with a seat minimum and an implementation phase on top, but none of that comes from the vendor, so treat it as hearsay until you have a written quote. Budget for the implementation and the renewal uplift as well as the licence.
Can you use Harvey without paying?
No. There is no free tier and no free trial. Harvey is sold exclusively through negotiated enterprise contracts, so solo practitioners and small firms cannot sign up directly and have to go through the sales team even for a demo. If you want to try legal AI on your own documents first, a cited document-QA tool with a free tier - Humata AI, for instance - is the cheaper way to find out whether the workflow suits you.
What are Harvey's closest competitors?
Legora is the closest like-for-like: an enterprise legal workspace built on Claude, and the one that publishes a starting figure, from $3,000 per seat per year on a 10-seat minimum. RealPact is the narrow option, voice agents that auto-fill real-estate brokerage contracts rather than general legal work. Humata AI is not really a competitor but is the honest cheaper answer if all you need is cited question-answering over your own documents, with a free tier and a $9.99 plan. Pick Harvey when the requirement is firm-wide workflow, matter context and a security review you can actually pass.
Does Harvey train on your client data?
No. The security page states Harvey contractually guarantees through its Platform Agreement that customer data remains the customer's and that inputs, outputs and uploaded documents are not used to train underlying models. Retention is configurable by the customer rather than fixed, data sovereignty is framed as a customer decision scoped in the contract, and the published certifications are SOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, GDPR and CCPA. Verified on the vendor's security page, 3 September 2026.
How do you get started with Harvey?
It begins with a sales conversation, not a signup form: request a demo, scope the seat count, sign an annual contract. Onboarding runs through implementation training, documentation and customer success sessions, with optional professional services for custom agents. Once live, lawyers use Harvey inside Word, Outlook and SharePoint as well as the web app, which is the part that decides whether adoption sticks - and the reason the platform ships a Command Center for administrators to watch usage.
Top Alternatives
HokAI guides covering Harvey
- What Harvey Is Actually For: Harvey just raised at an $11B valuation and serves 200,000+ lawyers. Here is who its four products are actually built for, and who cannot buy it at any price.
- Harvey vs Legora: The 35-Hour Onboarding vs the 6-Month Build: Harvey and Legora are both raising toward a $25.5B valuation in August 2026. What differs: a 6-9 month build versus 35 hours of onboarding, and seat cost.
- The Best AI for Research in 2026 Depends on the Job: Consensus, Gemini Notebook, Harvey and OpenEvidence each solve a different research problem in 2026. Here is how to match the right AI to your specific job.
- The Top AI Tools in 2026: How to Actually Choose One: ChatGPT, Claude, Cursor, Gemini, and Perplexity all cap out near $200 a month in 2026. A practical framework for matching the right AI tool to your actual job.