What is Harvey AI? Cost, uses and fit for small firms

Harvey is an AI platform for lawyers and professional services that works on legal documents and research: drafting, reviewing, analysing documents in bulk and answering legal, regulatory and tax questions. It is sold to firms and in-house teams on request, with no public price list. It suits firms with heavy document work; it does not handle calls, enquiries or client updates.

We are not Harvey and we do not compete with it. We build for the front of a law firm, the calls and the client relationship, which is a different job. That gives us a reason to explain Harvey accurately rather than talk it down: partners regularly ask us whether they need both, and the honest answer depends on what the firm's problem is. This piece sits in our guide to AI for UK law firms.

What does Harvey AI do?

Going by Harvey's own website, as of September 2026 the platform is a set of connected products rather than a single chatbot. The main ones it lists are:

• Agents: described by Harvey as purpose-built agents that carry out complex legal work end to end.
• Vault: a place to store, organise and bulk-analyse legal documents.
• Knowledge: research across legal, regulatory and tax questions.
• Spaces: shared workspaces for legal teams across organisations.
• Contract Intelligence: insight from contracts, to support negotiation and review.
• Harvey Mobile, integrations with tools lawyers already use, and an analytics layer Harvey calls Command Center.

Harvey also markets solutions by practice type (litigation, transactional, innovation teams) and by firm size, including a page for mid-sized firms. It publicly states that it raised $550 million at a $15.5 billion valuation, which gives a sense of the scale of investment behind it.

Put simply: Harvey operates on the artefacts of legal work. Documents in, analysis and draft work product out. That is a large and valuable category, and it is the one most people mean when they say 'legal AI'.

How much does Harvey AI cost?

Harvey does not publish its pricing. Its website offers a demo request rather than a price page. Various third-party articles circulate figures for per-seat costs and minimum commitments, but Harvey does not confirm them, and we have not seen them on any Harvey page, so we do not repeat them here. If price matters to your decision, the only reliable figure is the one in a written quote to your firm.

When you do get a quote, ask the questions that decide the real cost over a contract rather than the headline:

• Is it priced per seat, per usage, per matter or some mix? What counts as a seat?
• Is there a minimum number of seats or a minimum term?
• What happens to the price at renewal? Is any increase capped in the contract?
• Which products are included, and which are add-ons?
• What does onboarding and training cost, and who provides it?
• How do you get your data out if you leave?

Is Harvey AI suitable for small law firms?

Harvey's own marketing includes mid-sized firms, and it does not advertise a price floor. Whether it suits a small UK firm is a question about your work, not about Harvey. Ask three things.

• Is document work your bottleneck? If your fee-earners lose most of their week to drafting, review and research on long documents, a document platform addresses that. If they lose it to the phone, the inbox and chasing, it does not.
• Do you have volume? Bulk analysis earns its keep on large sets of documents. A high street firm doing conveyancing, wills and family work may not have the kind of volume the product is built around.
• Do you have someone to own it? Any serious AI tool needs a person who sets it up, trains colleagues and checks output. In a five-person firm that is a partner's time.

Our guide to AI for small law firms sets out a scoring method for choosing a first use. For many small firms, the first thing worth fixing is not document work at all.

There is also a question of timing. A firm that has not yet agreed which AI tools staff may use, on which accounts, and who checks their output is not ready to roll out any platform, however capable. Settle the policy first. It costs nothing, it takes an afternoon, and it means that when you do buy a document tool, the rules for using it are already in place rather than written in a hurry after something has gone wrong.

Is Harvey AI safe to use for UK legal work?

Harvey publishes material on its security approach, and any firm considering it should read it and ask for the supporting documents. The questions are the same for every AI vendor: where client data is processed and stored, whether it is used to train models, how long it is kept, who at the vendor can access it, and what certifications the vendor holds. Take the answers to your DPO and COLP.

The professional obligation does not move to the vendor. The Divisional Court's judgment in Ayinde v London Borough of Haringey [2025] EWHC 1383 (Admin) said lawyers who use AI for research must check its output against authoritative sources, and that the duty applies whether they used the tool themselves or relied on someone who did. That applies to every research tool, however good. See AI hallucinations and law firms for a checking routine.

What are the alternatives to Harvey AI?

Within document AI, the names most often mentioned alongside Harvey include Legora and CoCounsel, plus AI features added to the research databases and practice management systems firms already pay for. Compare them on the same questions: what they do with your documents, how they show sources, and what they cost at your volume. Our UK law firm software buying guide has a wider set of questions for any vendor.

Outside document AI there is a different set of tools altogether, and this is where the confusion usually sits. Tools that answer the phone, qualify enquiries, cover fee-earners and keep clients updated are not alternatives to Harvey. They are a different category solving a different problem.

How should a firm test Harvey, or any document AI, before buying?

The demo will be run on material chosen to show the product well. A fair test uses your material, under your conditions, with the results checked by the people who would rely on them. This is the method we would use for any document tool, Harvey included.

• Pick three real tasks from the last quarter where you already know the right answer: a review you completed, a research question you resolved, a document set you summarised. Anonymise them if the vendor's terms for a trial require it.
• Run each task through the tool and have the fee-earner who did the original work mark the output: what it got right, what it missed, and anything it stated that was wrong.
• Count the checking time, not just the generation time. A draft produced in two minutes that takes forty to verify has saved less than it appears to.
• Test the edge. Ask a question the material does not answer, and see whether the tool says so or fills the gap.
• Check every citation it produces against the official source, and record how many would have gone into a document unchanged had nobody looked.
• Ask the vendor, in writing, where your trial data went, whether it was retained, and whether it was used for anything beyond your session.

What you are measuring is not whether the tool is clever. It is whether, after checking, your fee-earners are faster and no less accurate. That is the only number that justifies a contract, and it is one only your firm can produce.

One trap worth naming: the partner who runs the trial is usually the most enthusiastic person in the firm, and the most forgiving reviewer. Ask a sceptical colleague to mark the output too. If both of them would sign off the result, the case is strong. If only the enthusiast would, you have learned something more useful than the demo told you.

Harvey vs front-door AI: which problem does your firm have?

A quick diagnostic we use with partners. Answer each with a rough number from the last month.

• How many hours did fee-earners spend reading, reviewing or drafting long documents?
• How many calls from new callers went unanswered or to voicemail?
• How many calls or emails were clients asking for an update?
• How long, on average, did a new enquiry wait for a reply?

A worked example, with numbers invented for illustration only. A hypothetical corporate team of twelve spends most of its time on due diligence and contract review, and misses very few calls because a switchboard covers them. That team has a document problem, and a tool like Harvey is aimed squarely at it. A hypothetical six-person high street firm handles modest documents but misses dozens of calls a month and fields constant update calls. That firm has a relationship problem, and a document platform will not reach it. Many firms have some of both, and there is nothing wrong with buying for each.

Where WasilX sits next to Harvey

We are not in the document-AI category. We do not read or summarise your documents and use no case law, so we do not draft legal work or answer legal questions. Cam, our AI Engagement Coordinator, takes calls on the firm's line at any hour, says it is AI, asks the firm's own qualifying questions, books the call back and sends the firm a summary. A personal assistant for each fee-earner covers their calls in court and drafts matter updates for the fee-earner to approve. For how a front-door system should behave, see AI chatbots for law firms. If that is your firm's gap, we are happy to talk.

Frequently asked questions

What is Harvey AI used for? Harvey is used for work on legal documents and research: drafting, reviewing and analysing documents, including in bulk, and researching legal, regulatory and tax questions, according to Harvey's own product pages.

Does Harvey AI publish its prices? No. Harvey offers demos and quotes on request and does not publish a price list. Third-party estimates exist, but Harvey does not confirm them.

Can a small UK law firm use Harvey? Harvey markets to mid-sized firms and does not state a minimum firm size publicly. Whether it fits depends on whether document work is the firm's bottleneck and whether someone can own the rollout.

Does Harvey answer client calls? No. Harvey is a platform for legal documents and research. Answering calls, qualifying enquiries and keeping clients updated is a separate category of tool.

Is Harvey AI's output always accurate? No AI research tool can be relied on without checking. The courts have said lawyers must verify authorities against authoritative sources whatever tool produced them.