← Back to AI Legal Lab
Insight
Generative AI for Legal Work

How legal workflows should change in the generative AI era

Hello, this is Legal Agent.

Using generative AI to draft faster does not necessarily reduce the time needed to complete a legal matter. General-purpose tools summarize contracts, analyze specific provisions, and draft revisions reasonably well. However, verifying requests, finding prior matter files, and entering tracked changes into working documents still consume substantial time.

Context lost outside the matter file

Pasting a clause into a browser works well enough for an isolated linguistic query. Practical contract review, however, takes place in a Word document containing counterparty redlines, comments, and internal commercial context, such as non-negotiable positions or plans for future business. Context omitted from a pasted fragment remains unavailable to the AI, meaning its output can appear sound while missing the actual operational needs of the transaction.

The operational core: active documents and matter history

Legal review advances through iterative edits in a Word file: adjusting text, recording redlines, explaining legal reasoning in margin notes, and responding to counterparty points, all while maintaining an appropriate tone for both external counterparts and internal teams. The matter folder also retains prior drafts and sticking points from comparable transactions, representing institutional memory that in-house counsel consults constantly. Where the task depends on this history, AI needs the relevant documents and context. Working with the matter files is one way to provide them.

AI groundwork and lawyer judgment

A liability cap should not be chosen by contract category alone; it depends on transaction size, commercial leverage, and past disputes. A practical arrangement relies on AI to handle preliminary tasks, such as highlighting issues, checking prior templates, and drafting candidate language. An experienced corporate attorney then refines that draft against company priorities. That combination, rather than raw AI output accepted without review, is what holds up in practice.

Separation between external comments and internal notes

An effective workflow does more than draft revisions quickly; it allocates each comment to the right place, audience, and tone. A note for the counterparty requesting a liability cap requires very different wording from an internal warning that uncapped exposure is disproportionate to deal value. Keeping external comments separate from internal risk assessments makes it easier for business teams to identify their next step.

Related articles

Articles connected to this topic.

Insight / 2026.07.15 Troubleshooting Generative AI: A Practical Checklist Insight / 2026.07.11 Why Human Review Is Essential Before Using Generative AI Output Insight / 2026.07.10 Verifying Statutory Citations and Sources in Generative AI Outputs
View AI Legal Lab articles