Harvey
Harvey is an enterprise generative-AI platform for legal and professional-services work. It handles research, drafting, summarisation and document review through a suite of Assistant, Vault, Knowledge, Agents and Contract Intelligence. Vault is the document engine: it stores, organises and bulk-analyses large sets and queries across up to 100,000 documents in a single vault, which suits discovery, due diligence and any matter that turns on reading a great deal at once.
Harvey publishes no pricing and has no self-serve path. Every deployment is a negotiated enterprise agreement priced on seat count, firm size, integrations and deployment scope, quoted through sales. Third-party reports and buyer accounts place annual contracts in the low-to-mid six figures, with reported per-seat rates differing by an order of magnitude between sources; Harvey confirms none of these figures publicly. The practical implication is that evaluating Harvey is a procurement exercise with a sales cycle and a negotiated commitment, not a tool you trial on a card.
Its edge over a general assistant is specialism and grounding: it is built for legal and professional-services material, and through Vault it works from a matter's own documents rather than open-web recall. That makes it a platform for teams producing legal work product at volume, not for the occasional legal question.
- Cost
- Enterprise
- Ease
- Model
- Hosted service
- Checked
- August 2026
Prices, plans and model versions change fast: this is a mid-2026 snapshot; check the tool's official site for the latest.
Best for
- Enterprise legal research, drafting, summarisation and document review
- Vault: bulk-analysing up to 100,000 documents per set
- Firm-wide deployments bought through an enterprise agreement
- A suite spanning Assistant, Vault, Knowledge, Agents and Contract Intelligence
- Querying across large legal document sets in one place
Less suited to
Harvey is not for the solo practitioner or small firm that wants to buy and use legal AI the same day: that reader is better served by a self-serve tool such as Spellbook. Nor is it general enterprise search across every connected system, where Glean goes wider, since Harvey's retrieval is scoped to the legal document sets and matter files it holds.
Where it does fit, treat rollout as a data-confidentiality and access-control decision. Harvey ingests confidential matter material at scale, so who may open which vault, what leaves the firm and how access is governed weigh as heavily as the platform choice itself; settle those with your risk and compliance owners before it becomes load-bearing.
Costs & data, in short
Enterprise-only: every deployment is an annual agreement priced on seat count, firm size, integrations and deployment scope, quoted through sales. There is no self-serve or credit-card path and no published pricing, so budgeting starts with a demo and a procurement conversation rather than a plan page.
Harvey holds confidential matter material at scale, so the data questions belong inside procurement rather than after it: agree vault access, retention and what may leave the firm with your risk and compliance owners as part of the agreement, and treat rollout as an access-control decision as much as a purchase.
The vendor does not publish a price list.
In practice
How Harvey is used, area by area.
Jobs
Legal & compliance
Harvey's fit for legal and compliance work starts with a constraint worth stating plainly: it is sold only through enterprise agreements, with minimum seat commitments and no self-serve path, so it suits a firm or department buying at scale rather than an individual testing an idea. Within that frame it covers the core professional-services tasks, research, drafting, summarisation and document review, and pairs them with Vault, which stores, organises and bulk-analyses large document sets and queries across up to 100,000 documents at once. The wider suite adds Knowledge, Agents and Contract Intelligence around the same matter files. Legal teams that can commit to a firm-wide rollout, and need to reason over large document sets, gain the most.
Example tasks
- Research legal questions and draft documents across professional-services work
- Review and summarise contracts and large matter files
- Bulk-analyse up to 100,000 documents in a single vault
- Query across organised document sets during discovery and due diligence
- Deploy Assistant, Knowledge, Agents and Contract Intelligence firm-wide
Limits
There is no self-serve entry and deployments carry minimum seat commitments, so a solo practitioner or small firm wanting to buy and start today is not the fit.
Pricing is unpublished and sales run through an enterprise agreement, so budgeting needs a demo and a procurement conversation, and the confidential matter material the platform ingests makes rollout a data-confidentiality decision as much as a tooling one.
Compares
| vs | Pick Harvey when | Pick the other when |
|---|---|---|
| SpellbookFull comparison → | Harvey is an enterprise platform bought through an agreement with minimum seat commitments, not a tool a smaller firm can sign up for and use the same day | you want self-serve legal AI a solo practitioner or small firm can actually buy without a sales process |
| LegoraFull comparison → | the confidential matter material Harvey ingests makes rollout a data-confidentiality decision as much as a procurement one, which is a different conversation from choosing a feature set | tabular review and precedent-grounded drafting are what the firm-level brief actually calls for |
| Gemini NotebookFull comparison → | Harvey deploys Assistant, Knowledge, Agents and Contract Intelligence across a firm rather than a notebook across a matter | the questions are about what these specific documents say, with the passage to prove it |
Tasks
Search & knowledge retrieval
For search and knowledge retrieval
For search and knowledge retrieval, Harvey's centre of gravity is Vault: it stores, organises and bulk-analyses large document sets and queries across up to 100,000 documents in a single vault, with Knowledge holding reference material the team relies on. This is retrieval scoped to legal matters, discovery, due diligence and any question that turns on reading across a large set, rather than general search over every connected system. Because it is enterprise-only, with no self-serve path and minimum seat commitments, adopting it for retrieval is a firm-level decision made through a sales process. Legal teams that need to query across large matter document sets, not their whole app estate, gain the most.
Example tasks
- Query across large document sets stored in Vault
- Bulk-analyse up to 100,000 documents in one vault
- Retrieve answers from organised matter files and knowledge
- Run discovery and due-diligence reads over large sets
- Search across drafting, research and contract material in one place
Limits
Search across the wider app estate, connected systems and everyday knowledge sources is not what this is; retrieval is scoped to the document sets and matter material held inside the platform.
Adoption is enterprise-only, with no self-serve path, unpublished pricing and minimum seat commitments, so standing up retrieval here is a firm-level procurement decision rather than a quick self-serve setup.
Compares
| vs | Pick Harvey when | Pick the other when |
|---|---|---|
| Glean | Harvey retrieves across legal document sets and matter files held in Vault rather than searching across every connected system in the company | the job is enterprise search over connected apps and knowledge sources rather than reasoning over a legal matter's documents |
| LegoraFull comparison → | standing retrieval up in Harvey is a firm-level procurement decision rather than a setup: enterprise-only adoption, no self-serve path and minimum seat commitments | the retrieval surface should be tabular, asking the same questions across hundreds of documents at once |
| Gemini NotebookFull comparison → | Harvey bulk-analyses up to 100,000 documents in a single vault, which is a different order of set entirely | the knowledge to search is a corpus you can assemble, and provenance matters |
Writing & research
In writing and research
In writing and research, Harvey covers the professional-services core: research, drafting, summarisation and document review, delivered through Assistant and organised alongside Knowledge. It is built for legal and professional work specifically rather than general writing, so the drafting and research assume matter files, contracts and filings as the material. Vault backs this by holding large document sets and querying across up to 100,000 documents, so a research task can draw on a whole matter rather than a handful of pasted pages. The catch is procurement: there is no self-serve entry, only an enterprise agreement with minimum seats, so this is a team decision. Legal teams that draft and research at volume and can commit firm-wide gain the most.
Example tasks
- Draft legal documents, memos and correspondence from a prompt
- Summarise contracts, filings and long matter files
- Research legal questions across professional-services subject matter
- Review documents and surface issues for a matter
- Organise research through the Knowledge and Assistant surfaces
Limits
General-purpose writing is not the ground here; the drafting and research assume legal and professional-services material and a matter behind it.
There is no self-serve signup and deployments carry minimum seat commitments with unpublished pricing, so adopting it for drafting alone means an enterprise agreement and a procurement conversation, not a same-day subscription.
Compares
| vs | Pick Harvey when | Pick the other when |
|---|---|---|
| SpellbookFull comparison → | Harvey approaches drafting and research as an enterprise platform across the full matter, sold through an agreement rather than a self-serve subscription | a smaller firm wants contract drafting and review it can buy and start using without an enterprise commitment |
| LegoraFull comparison → | Harvey's drafting and research assume legal and professional-services material with a matter behind it, so general-purpose writing is simply not the ground | the writing must track the firm's own precedents and the positions have to be defensible from them |
| Gemini NotebookFull comparison → | Harvey drafts legal documents, memos and correspondence from a prompt rather than only reporting what the sources say | the research corpus is defined and answers must trace to it |
Where to start
Not sure what to adopt first?
Five quick questions about your job, task and constraints. We'll suggest your top three tools, plus the one to try first.
Alternatives
Same category, different strengths.
Appears in these stacks
Curated combinations this tool is part of.
Common questions
Is Harvey free?
No: Harvey is enterprise software, priced per organisation.
Where does Harvey fit best?
Harvey fits best in Legal & compliance and Writing & research; see its practice notes for how.
Before sharing confidential or personal data, check this tool's data-governance and training policies. They differ between providers and can change.
Last checked: August 2026