Research, review, summarising and first drafts. The fastest moving category here, and the one where claims need the most checking.
Legal AI assistants use large language models for research, document review, summarising and first drafts. Harvey, Legora, CoCounsel, Luminance and Spellbook are the most widely adopted; check where client data is stored, whether it is retained, and whether it is used for training before you upload anything.
Updated 30 October 2026. 10 products, listed alphabetically.
Legal AI assistants: 10 products
Listed alphabetically. A listing is not a recommendation, and sponsored placements are labelled. Links go straight to the vendor.
AI legal assistant from Thomson Reuters, built on the Casetext acquisition.
Drafting assistance that surfaces definitions and cross-references while you read.
AI platform for in-house legal teams, covering contracts, compliance and legal operations work.
AI for personal injury firms: demand letters, medical record review and case preparation.
AI agents that handle routine legal requests and contract work for in-house teams.
AI assistant for legal work, adopted by a number of large firms and legal departments.
Collaborative AI workspace for lawyers, used across research, review and drafting.
Document analysis and contract negotiation, with roots in diligence review.
Contract drafting and review assistance inside Microsoft Word.
Consumer-facing AI legal assistant that answers legal questions and connects people with lawyers.
Is it safe to put client files into these?
It depends entirely on where the file goes and what happens to it. A consumer chatbot is not a safe place for privileged material. A properly configured legal product, or a private setup on infrastructure you control, is a different proposition. Ask three questions of every vendor: which jurisdiction the data sits in, whether anything is retained after the session, and whether your content is ever used for training. Get the answers in writing.
What are these genuinely good at today?
First drafts, summarising long documents, finding things across a large set of files, and first-pass review where a human checks the output. They are least reliable where the answer depends on facts nobody has given them, and where a confident wrong answer is expensive. Treat output as a junior's work that still needs reviewing, because that is what it is.
How this directory works. Products are selected and described by the Legal Desire editorial team. A listing is not a recommendation, and descriptions say what a product is for rather than how good it is, because the honest answer depends on your firm.
The directory carries both editorial entries and commercial placements. Anything commercial is labelled as sponsored on the page and in the structured data, so you always know which is which. Vendors can apply for a listing or a sponsorship.
For vendors
Want your product in front of this audience?
This page is read by the lawyers and legal teams choosing legal AI assistants. Legal Desire has published legal industry news and analysis for 14 years, read by lawyers, general counsel, firm partners and legal operations teams across the United States, the United Kingdom, Europe and the Gulf.