You can watch replay of session or download the resources by clicking the button below.

About this masterclass

The next instalment of our masterclass series focuses on how risk and compliance teams can build AI agents for regulatory change management. Last time we traced a regulatory change through to the policies it impacts. This time we take a wider view of how to design and implement AI workflows across regulatory change management with humans in the loop at the appropriate stages.

By joining this masterclass risk and compliance teams will learn how to design AI workflows with humans in the loop. The session is run in partnership with the ICA and designed for practitioners implementing AI into production.

The regulatory change workflow design splits into three stages. Each stage can fail differently without oversight and governance, so each section needs a different approach to workflow design and keeping human expertise in the loop to safeguard output:

1. Obligation extraction. The agent reads the publication and extracts the relevant obligations, filtering alerts from impactful change to commentary for monitoring.

2. Impact assessment. Learn how AI agent workflows can map obligations to policy documents or procedures and controls that already speak to it, at clause level rather than document level.

3. Suggested fixes. Learn how an AI agent can draft the remediation, written against your firm's own language, structure and existing policy set rather than a generic template.

What you will learn:

How to design a regulatory change management workflow as a sequence of AI agent steps with deliberate human checkpoints, rather than one automated pipeline

Extracting discrete obligations from regulatory text, and which judgements have to stay with a person

Mapping obligations to internal documents like policies or controls at clause level

Generating remediation suggestions grounded in a firm's unique policies, controls and risk appetite

What a reviewer should actually be checking at each step, and how to keep that review proportionate

Capturing sign-off and reasoning so evidence of coverage stays audit-ready between formal reviews