Human Accountable for The Loop

AI systems are moving beyond producing answers. They can route work, apply rules, communicate with customers, generate documents and trigger actions across other systems. As their authority grows, the familiar promise of keeping a “human in the loop” becomes less convincing.

A person may be present without having the information, time, authority or technical means to change the outcome.

Human Accountable for the Loop presents a practical framework for governing AI-enabled workflows. It asks eight connected questions: Who owns the outcome? What authority has been delegated? Which limits are technically enforced? How can people escalate concerns? Can important events be reconstructed? What is monitored in production? When must approval be reconsidered? Where do liability and remedy sit?

The book develops these questions through real cases involving automated decisions, professional services, autonomous vehicles, public administration and AI products. It then applies the complete framework to a worked contracting example and provides a practical workbook for assessing live workflows.

I wrote this book for leaders, lawyers, risk professionals, product teams, engineers and operational owners who need to decide when automation may act, and how their organisation remains answerable when it does.