Updated 23 October 2026.
AI can assist legal teams with defined tasks, but a universal claim that it replaces 40% of legal work is not supported by the evidence cited here. Results depend on the task, tool, documents and professional review. Savings should be measured in a controlled pilot.
Choose a bounded workflow
Start with document organization, first-pass summaries or a checklist comparison. Define the input, expected output and who reviews the result. For client matters, use tools and data-handling arrangements approved by the responsible firm.
Keep professional responsibility with the lawyer
The American Bar Association’s Formal Opinion 512, issued 29 July 2024, discusses competence, confidentiality, communication, supervision, candor and reasonable fees when lawyers use generative AI. Generated text and citations need verification. Applicable jurisdictional rules and client requirements remain relevant.
Measure the whole process
Compare similar tasks with and without the tool. Record preparation, review and correction time, material errors and total cost. A faster first draft may require additional checking. Report the sample size, dates, exclusions and review method alongside any result; do not generalize a pilot to all legal work.
An illustrative review workflow
For an internal, non-confidential sample document, a team could prepare a human reference summary, generate an AI draft, and compare omissions and inaccuracies before considering a wider pilot. This is a suggested process, not a reported Menteso client result.
Source and next steps
ABA Formal Opinion 512: Generative Artificial Intelligence Tools (PDF).
Explore AI agent development or remote paralegal support to discuss a scoped workflow. This article provides general information and does not replace advice from the responsible legal professional.
