InsightGovernance
A policy is not an audit trail: what operational AI governance actually requires
Written expectations describe intent. Assurance requires a record of what a system did, who reviewed it and on what basis.
Coming soonInsights
AI governance for people who have to make it work.
Working notes on the practical side of governed AI: where oversight breaks down, what an audit trail has to contain, and how firms keep adoption moving without accumulating exposure.
Editorial note
Articles below are planned pieces, published as they are written. We do not publish client work, named examples or statistics we cannot evidence.
InsightImplementation
Why human oversight fails when it is added after deployment
Review steps bolted on to a live workflow become optional in practice. Oversight has to be a designed part of the process.
Coming soonInsightRisk & Assurance
The questions your insurer may eventually ask about AI
Professional indemnity conversations are moving from whether a firm uses AI to how its use is controlled and evidenced.
Coming soonInsightLaw
Supervision, delegation and AI: an old professional duty in a new setting
The supervisory question a firm answers about a junior fee earner is the same question it now has to answer about a model.
Coming soonInsightAccountancy
Review controls when part of the work is machine-generated
Quality review assumes a reviewable trail. Generated outputs need the same traceability as any other working paper.
Coming soonInsightGovernance
Minimum necessary governance: controlling risk without stalling adoption
Over-engineered governance is abandoned. The objective is the smallest control set that makes a use case defensible.
Coming soonNext step
Would rather discuss it than read about it?
An AI Risk & Readiness Audit applies these questions directly to your firm, with findings written for your partnership.