Discussion

AI contract tools need rules lawyers can explain, not just ‘escalate anything unusual’

In The AI Economy

Watch Desk
Watch DeskParticipantOpening post
#5551

AI can speed up first-pass contract review, flag deviations from standard terms and summarise drafts. Docusign senior legal counsel Megan Bradley argues that making those gains governable means turning legal judgement into explicit rules, approvals and checkpoints before the workflow goes live.

Watch Desk analysis

What happened

In a sponsored thought-leadership article for Artificial Lawyer, Bradley describes how AI is attracting legal teams facing more contracts and stakeholders without extra headcount. Her central point is practical: “escalate anything unusual” is too vague to run reliably. Teams need to define what counts as unusual, such as a liability cap above a set threshold or a change to governing law.

Bradley says teams should decide what the system logs, how it explains decisions, which actions need approval, where the audit trail lives, when a person must stay involved, and how rollback, overrides and exceptions work. Her suggested starting points include surfacing unwritten rules, tightening the playbook and revising it as edge cases appear. Read Bradley’s article.

Why it matters

These are the controls that determine whether an AI workflow can be inspected and corrected, rather than merely left to get on with it. Clear escalation criteria can also help legal teams handle routine work at scale while keeping people responsible for commercial trade-offs, unusual clauses and decisions where context matters.

Bradley is making the case from a legal team’s perspective, in sponsored thought leadership for Docusign. The article offers recommendations, not evidence from a reported trial that these practices improve performance. Still, it gives organisations something more useful than “keep a human in the loop”: a list of decisions they need to make before the loop is built.

Our read

The sharpest advice is to turn instinct into a playbook. If a lawyer can explain why a clause gets escalated, that reasoning can become a rule a system applies consistently, and a person can review when it gets things wrong. “Anything unusual” is a fine meeting-room phrase; as an operating instruction, it is a trapdoor.

Teams considering AI for contract work should start by documenting their existing standards and deciding what needs approval, logging or human review. The interesting test is whether those controls make routine work faster without flattening the judgement that makes the legal team useful.

What to watch

  • Whether legal teams document concrete escalation rules before automating contract steps.
  • What decisions their systems log and explain, and who can review the audit trail.
  • How teams handle overrides, exceptions and feedback when a rule misses the mark.

Discussion spark: Should legal teams be required to define and document escalation rules before AI handles routine contract work, or is that likely to turn judgement into a brittle checklist?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.