Harvey AI, and the alternatives

Harvey is a legal AI platform. It describes itself as built for "the world's top law firms and in-house legal teams", and it is sold as a set of modules around a shared core: agents for multi-step tasks, a vault for documents, legal research, shared workspaces, and add-ins for Word and Outlook.

It runs on models from several providers (Anthropic, OpenAI and Google), with an Auto setting that picks one per task. Lawyers use it in a web app, hosted in the US, the EU or Australia, or from inside Word and Outlook. It connects to iManage and NetDocuments, the document systems law firms run on.

The useful way to understand Harvey is by what it is organised around: the law-firm matter. Work arrives as a matter, a lawyer opens it, and Harvey's modules help that lawyer research, draft and review faster. That shape decides who it suits.

1. Agents. Multi-step tasks, carried out on a lawyer's instruction.
2. Vault. Store documents and analyse them in bulk.
3. Knowledge. Legal, regulatory and tax research.
4. Spaces. Shared workspaces for working with co-counsel, with ethical walls between matters.
5. Contract Intelligence. Analysis across a set of contracts.
6. Horizon Scanning. Tracking of regulatory change.
7. Command Center. Usage analytics across the licence.
8. Intake and Legal Work Orchestration. Marked new on Harvey's site, and aimed at in-house teams.

Harvey doesn't publish prices. It sells by quote, and reviews report minimum licence counts and 12-month contracts. LawXYAI estimates about $1,200 per seat per month with a 20-seat minimum, about $288,000 a year to start. Vaquill, itself a legal AI vendor, estimates $1,200 to $2,000 or more per seat. Both are estimates.

On security, Harvey lists SOC 2 Type II, ISO 27001, ISO 27701, ISO 42001 and AIUC-1, with SAML single sign-on, audit logs and IP allow-listing.

The legal AI pricing guide sets Harvey's estimates beside Legora's, Spellbook's and Claren's published plans, with a worked example for a five-person in-house team.

Search for Harvey reviews and most of what comes back is about price, because Harvey doesn't publish one. LawXYAI's review asks whether it is worth $288,000 a year for legal teams. Other guides estimate per-seat costs, seat minimums and contract terms.

Read the bylines. Several of the guides that rank for Harvey's pricing and reviews are written by other legal AI companies, and many repeat each other's figures. Searches for Harvey's competitors return directories and analyst lists with little guidance for a buyer.

What the reviews rarely ask is the question that matters most for an in-house team: once it is bought, who does the work? Harvey's modules are tools. A lawyer opens each one, gives it the task, checks the result and files it. For a firm with many lawyers, that is the point. For an in-house team with more requests than lawyers, it is the problem.

Harvey suits a law firm that wants one platform for research, drafting and document review across many practice areas, with ethical walls between matters and space to work with co-counsel. It suits firms that already run on iManage or NetDocuments, and that have enough lawyers to fill the reported 20-seat minimum.

It is less suited to an in-house team. In-house work doesn't arrive as a matter. It arrives as a message from sales about a customer's paper, a renewal nobody flagged, a supplier asking for the DPA back by Friday, and twenty NDAs a week from across the business. Somebody has to notice each request, pick it up, do it against the team's positions, send it back and file it. It is not suited to in-house intake: even its intake module hands each request back to a lawyer to drive.

1. Legora. A legal workspace built around an Agent, Tabular Review (one grid that pulls the same answers from many documents) and Workflows, with research, Word and Outlook add-ins, a mobile app and a client Portal. Sold to firms and in-house teams by quote. See Harvey vs Legora.
2. Spellbook. Contract drafting and review inside Microsoft Word: redlines, playbooks, market comparison and ACM for intake and renewals. Contracts only, with no litigation or research. Sold by quote, with a 7-day free trial.
3. Claude for Legal. Anthropic's twelve practice-area plugins and twenty-three connectors on its general Claude model, with tracked-change redlines in Word. Each lawyer sets it up in their own Claude. See Claude for legal.
4. Claren. AI employees for in-house legal teams. Envoys take requests from Slack and the Claren app, review against the team's playbook, redline in Word and keep the team's contract records. Published per-seat plans.

Every tool in that table, Harvey included, makes a lawyer faster. Claren is built the other way round. It gives the team envoys: AI employees, each with its own computer and its own memory, that take a request, do the work and hand it back finished.

Envoys take requests from the Claren app or straight from Slack, where the business already asks. They review every contract against the team's playbook and hand back tracked-change redlines in Word, with a comment on every change, fallback positions applied and a summary of what's off-playbook. They file each contract and its terms in the team's datastore, and build the trackers and dashboards the team needs on top of it.

Routines run on a schedule or when a Slack message lands: renewals checked each morning, every contract in a channel read and triaged. Envoys pass work to each other, and ask before anything leaves the team. The playbook goes in once, and every envoy follows it.

The results are measured in hours. Contracts come back redlined, data rooms come back indexed, renewals get flagged before they roll. The team gets hours back every week, from the first month.

1. Are you a firm, or a legal team inside a business? Harvey and Legora are shaped around law-firm matters. An in-house team should start from how its own work arrives.
2. Do you need a faster lawyer, or the work done? Harvey, Legora, Spellbook and Claude each make the lawyer at the keyboard faster. Claren's envoys do the work and hand it back for sign-off.
3. Where do requests come from? If they come from the business, in Slack and email, you want something that picks them up there. Claren's envoys do.
4. Who keeps the records? With a tool, the filing is yours. Claren's envoys file every contract and its terms as they go, and build your trackers on top.

Most in-house teams start with the work that repeats: NDAs from sales, supplier contracts, the DPA a customer wants back by Friday. Put the team's playbook in once, point an envoy at the channel where those requests land, and count the hours that come back.

Claren's plans are published, from a free plan for one member to Team for three or more, with usage included. Teams that need zero data retention and single sign-on move to Enterprise.