Legal AI use cases fall into five families. Knowing the families helps you compare options and pick a starting point.
- Read and summarize: contract summaries, case and transcript summaries, regulatory update digests, board paper summaries. Low effort, fast value, and the output is checked by the lawyer who reads it.
- Find and extract: clause identification, obligation and date extraction, due diligence tables, playbook checks against your standard positions.
- Draft and revise: first drafts of NDAs, policies, memos, emails and contract mark-ups against a playbook.
- Research and answer: legal research tools grounded in case law and legislation; internal Q&A over your own policies and templates.
- Review at scale: e-discovery and investigations, where AI ranks or classifies large document sets for lawyers to review.
There are also operational uses: intake triage, matter routing, outside-counsel bill review, transcription of meetings and interviews, and translation.
Where legal departments actually are
In The General Counsel Report 2026, the most common generative AI tasks were summarization (83% using or experimenting), identifying contract clauses (63%), audio and video transcription (53%), analysis of foreign-language materials (40%) and first-pass review (37%). The pattern is clear: teams start with reading and finding, then move to drafting and review at scale.
How to compare use cases
Score each candidate 1–5 on:
- Value: hours saved, speed, quality or risk reduced.
- Effort: data to prepare, integration, change for the team.
- Risk: sensitivity of the data, consequence of an error, who sees the output.
Start where value is high and risk is low, with a human checkpoint and one KPI. The use-case catalog in this hub scores the common ones for you.