AI for Legal Workflows

Lawyers, in-house counsel and paralegals.

Use AI across drafting, review and research safely — with a clear view of confidentiality and accuracy risk.

AI is already producing made-up cases and false citations that have embarrassed legal teams, and client confidentiality is easy to breach without realising it. As clients and colleagues expect faster turnaround, the temptation to lean on AI without checks grows. Clear rules now let your team move quickly while protecting privilege and accuracy.

• Tell where AI genuinely helps legal work from the specific places it becomes risky
• Use AI to draft, summarise and review documents without losing control of quality
• Research with AI while avoiding the made-up cases and false citations it can produce
• Apply clear, simple rules that protect confidentiality and privilege
• Check AI output thoroughly enough that you can put your name to it
• Leave with a starter AI usage policy and a prompt library to build on

1. Where AI helps legal work — and where it's risky

   • What today's AI tools can and can't do for legal tasks, in plain terms
   • General chatbots versus purpose-built legal tools, and why the difference matters
   • The points in legal work where AI helps versus where it must not be trusted alone
   • The duty to stay competent in the tools you use
   • Setting expectations before a team starts using AI

2. Drafting, summarising and reviewing with AI

   • Drafting clauses, memos and first-pass documents from a clear brief
   • Summarising long documents, bundles and correspondence
   • Reviewing a contract by asking questions of the document
   • Marking up against a standard position or playbook
   • Keeping the lawyer, not the tool, in control of the final product

3. Writing prompts for legal tasks

   • The anatomy of a good legal prompt: role, context, task, constraints, format
   • Asking the tool to reason step by step for better-reasoned output
   • Prompt patterns for drafting, review and research
   • Iterating and refining when the first answer falls short
   • Building reusable prompt libraries and document playbooks

4. Research without the made-up cases

   • Why open chatbots are dangerous for legal research
   • How AI invents authoritative-looking but fake citations and case law
   • Using tools grounded in real, curated legal sources instead of guesswork
   • Requiring AI to point to specific authority you can open and read
   • A verification step you never skip before relying on a result

5. Confidentiality and privilege: clear rules for AI use

   • How client information can leak into outside tools without you noticing
   • The risk to privilege and confidentiality when data leaves your control
   • Reading a tool's data-retention and training terms before you trust it
   • When client consent is needed before AI touches their matter
   • A short set of rules the whole team can follow

6. Checking AI output you can put your name to

   • Treating every AI draft as unverified until checked
   • A practical review routine for AI-assisted drafting and research
   • Independently confirming every citation and quoted authority
   • Spotting invented facts, wrong dates and subtly wrong reasoning
   • Owning the final work product as your own

7. A simple AI usage policy and prompt library

   • A starter usage policy covering approved tools, data and checks
   • The confidentiality and verification rules on a single page
   • A prompt library for drafting, summarising and review
   • Guidance on when AI is and isn't appropriate for a task
   • A policy and library your team can adopt and extend