Harvey

Harvey is the enterprise AI platform built specifically for law firms and in-house legal teams. The product suite covers Assistant (legal chat), Vault (secure document storage and retrieval-grounded drafting), Workflows (pre-built legal task agents), Knowledge (firm-wide regulatory research), and Agents (end-to-end execution of multi-step legal work). All of it runs on models customized for legal use, with OpenAI as an early partner and seed investor.

The company was founded in 2022 by Winston Weinberg, a former O’Melveny litigator, and Gabriel Pereyra, a former Google DeepMind and Meta researcher. By January 2026 Harvey hit $190M in annual recurring revenue. In March 2026 it raised $200M at an $11B valuation led by GIC and Sequoia, bringing total funding above $1B.

As of April 2026, Harvey lists the majority of the AmLaw 100, over 500 in-house legal teams, and 50 asset managers across 60 countries as customers. More than 25,000 custom agents run on the platform.

Recent developments

• April 30, 2026: Legora added $50M from NVentures and Atlassian, reaching a $5.6B valuation. Harvey remains the larger legal-AI platform by reported valuation, but Legora is now the clearest scaled challenger.

System Verdict

Pick Harvey if the firm is an AmLaw 100 or equivalent and needs legal-domain AI with retrieval grounded on firm documents. Vault keeps drafting tied to the firm’s own precedent and client files. Workflows covers the recurring heavy lifting: M&A due diligence, contract review, fund formation. Shared Spaces coordinates work across in-house teams and external counsel. The $11B valuation, $190M ARR, and 60+ AmLaw 100 adoption are the clearest signals that the category leader has been chosen.

Skip it if the firm is too small to justify a six-figure contract or needs a general-purpose assistant. Harvey has no published self-serve tier. Reported ACV sits around $1,000 to $1,200 per lawyer per month before any custom-model work, which prices out most boutiques and solo practitioners. For general drafting, ChatGPT or Claude cost a fraction.

Who pays which tier: everyone is Enterprise. The only decision is annual contract value, which scales with seat count, Vault storage, workflow customization, and whether the firm commissions a custom fine-tuned model.

Key Facts

What it actually is

Harvey is a legal-domain AI platform, not a chatbot wrapper. Four things distinguish it from using ChatGPT Enterprise or Claude for Work:

First, Vault retrieves from a firm’s own document store (precedent, client files, contracts) and grounds drafting on those documents. That is the core deliverable law firms pay for. Generic LLMs hallucinate case law; Vault pulls from files the firm already owns.

Second, Workflows are pre-built task agents tuned for legal work. M&A due diligence, contract review, fund formation, litigation research, regulatory memos. Over 25,000 custom agents run on Harvey today, with long-horizon agents handling multi-step processes over days.

Third, Shared Spaces lets a firm’s lawyers and external counsel (or in-house teams plus their outside firms) collaborate on the same matter inside Harvey with permissions and audit trails.

Fourth, custom models. Large firms can commission Harvey to fine-tune a model on firm-specific precedent and style. That is the high-ACV tier.

When to pick Harvey

• Large law firm or in-house legal team. AmLaw 100 scale is the baseline. 500+ in-house teams also use it. If the firm has a dedicated innovation or legal-ops budget, Harvey is the default first call.
• Document-grounded drafting. Vault is the differentiator. Any workflow where outputs must cite the firm’s own precedent or a client’s own contracts is Harvey territory.
• Recurring heavy workflows. M&A due diligence, fund formation, contract review at volume. Workflow agents beat bespoke prompt engineering.
• Cross-firm coordination. Shared Spaces handles the in-house plus outside-counsel pattern with proper governance.
• Custom fine-tuned models. Firms with a distinctive drafting style or practice area can commission a custom model trained on their corpus.

When to pick something else

• Solo or small-firm practice: ChatGPT or Claude at $20/month each. Good enough for non-privileged research and first drafts. No six-figure contract.
• General-purpose legal research outside a firm: ChatGPT Plus plus a Westlaw or Lexis subscription. Harvey does not replace primary legal research tools.
• On-prem or air-gapped deployment: Harvey is cloud-native. Firms with absolute data-sovereignty requirements should look at Reka or other on-prem-capable enterprise LLM vendors.
• Enterprise content generation outside legal: Writer for marketing, comms, and knowledge work. Writer runs its own Palmyra models and is not legal-specialized.
• Contract-review-only use case at lower price: narrower point tools (not in our catalog at this tier) undercut Harvey on single-purpose contract analytics. Harvey wins on suite breadth, not per-feature cost.

Pricing

Pricing is enterprise contact-sales only. Harvey publishes no list price on harvey.ai.

Third-party coverage reports annual contract values in the range of $1,000 to $1,200 per lawyer per month, scaling by seat count, Vault storage, workflow customization, and whether a custom model is commissioned. Industry reporting has also cited deal sizes from mid-six figures to over $500K ACV for large firms.

There is no free tier and no public free trial. Pilots run through enterprise sales.

Prices verified 2026-04-18 via the Harvey website and CNBC coverage of the March 2026 funding round.

What this buys: Assistant, Vault, Knowledge, Workflows, Agents, Shared Spaces, SOC 2 Type II and ISO 27001 compliance, and access to multiple underlying frontier models (OpenAI, Anthropic, Google) plus any custom fine-tunes.

Against the alternatives

Failure modes

• No public pricing. Every deal is bespoke. Comparing value across firms, or even across renewal cycles, requires back-channel intelligence.
• Price floor excludes small firms. At roughly $1,000+ per lawyer per month, Harvey is not a realistic purchase for most boutiques, regional firms, or solo practices.
• Cloud-only. Firms with strict data-residency or air-gap requirements cannot deploy Harvey on-prem.
• Hallucination risk remains. Vault grounding materially reduces hallucinations but does not eliminate them. Citations and case references still need lawyer review before filing.
• OpenAI dependency. Harvey’s deepest model partnership is with OpenAI, which is also a seed investor. Model-level outages or pricing shifts at OpenAI flow through to Harvey workflows.
• Opaque model routing. Which underlying model (OpenAI, Anthropic, Google, or custom) handles a given task is not always surfaced to the end user.
• Long procurement cycle. AmLaw-scale procurement, security review, and pilot design can run six to twelve months before go-live.
• Category competition intensifying. New legal-AI entrants are appearing monthly. Harvey’s lead is real but not unassailable on any single feature.

Methodology

This page was produced by the aipedia.wiki editorial pipeline, an automated system that ingests vendor documentation, verifies pricing and model details against primary sources, and generates the editorial analysis you are reading. No individual human wrote this review. Scoring follows the four-dimension rubric at /about/scoring/ (Utility × Value × Moat × Longevity, unweighted average). Last verified 2026-04-18 against the Harvey website, the Harvey blog post on the $11B funding round, CNBC coverage, and Reuters reporting on the March 2026 raise.

FAQ

What does Harvey cost? Enterprise contracts only. Harvey publishes no list price. Third-party reporting puts ACV roughly between $1,000 and $1,200 per lawyer per month, with large-firm deals running into the mid-six figures annually. There is no free trial.

Which underlying AI models does Harvey use? Harvey runs on a mix of frontier models (OpenAI, Anthropic, Google) plus custom fine-tunes. OpenAI was the seed investor and remains the deepest partner, though Harvey is not exclusively GPT-based.

Who uses Harvey? The majority of the AmLaw 100, over 500 in-house legal teams, and 50 asset management firms. Named clients include A&O Shearman, Reed Smith, Vinson & Elkins, O’Melveny, CMS, Dentons, BakerHostetler, DLA Piper, Deutsche Telekom, HSBC, and NBCUniversal.

What is Vault? A secure document store plus retrieval-grounded drafting. Upload firm precedent, client contracts, and matter files. Vault grounds LLM output on those documents, which reduces hallucinations relative to ungrounded chat.

What are Harvey Workflows? Pre-built task agents tuned for recurring legal work: M&A due diligence, contract review, fund formation, litigation research, regulatory memos. Over 25,000 custom agents run on Harvey today, including long-horizon agents that span multi-day processes.

Is Harvey secure enough for privileged client data? Harvey holds SOC 2 Type II and ISO 27001 certifications, and is GDPR and CCPA compliant. Major law firms including A&O Shearman, Reed Smith, and Vinson & Elkins have cleared it through procurement. On-prem deployment is not offered; the platform is cloud-native.

Related

• Category: AI Research · AI Writing · AI Automation

Embed this score on your site Free. Links back.

HTML Copy

<a href="https://aipedia.wiki/tools/harvey/" target="_blank" rel="noopener"><img src="https://aipedia.wiki/badges/harvey.svg" alt="Harvey on aipedia.wiki" width="260" height="72" /></a>

Markdown Copy

[![Harvey on aipedia.wiki](https://aipedia.wiki/badges/harvey.svg)](https://aipedia.wiki/tools/harvey/)

Badge value auto-updates if the editorial score changes. Attribution via the link is required.