\cleardoublepage \part{The Operator's Own AI} \setbookpart{The Operator's Own AI}

Part IV: The Operator's Own AI

The operator's AI is part of the estate it operates. Remediation agents, observability pipelines, runbook libraries and configuration maps hold the broadest credentials and the best map of the organisation. Part IV treats the agent team, the observer principle and the defence of the operator's AI.

Chapter 17 is the book's central design rule and stages a debate.

Chapters in this part:

• Chapter 16, An AI Agent Team for Operations, describes the agent team, its skills, mandates and containment (all five functions).
• Chapter 17, The Observer Principle, states the observer principle and stages a debate (O1 Observe, O4 Restore).
• Chapter 18, Defending the Operator's AI, treats threats to operations agents and the control plane as Tier 0 (O3 Change, O4 Restore).

Functions in this part. Part IV concerns all five functions through the agent team; the observer principle concerns O1 Observe and O4 Restore, and the defence of the operator's AI concerns O3 Change and O4 Restore.

The thread from the series. The method of conducting an agent team is that of The Transformed Security Architect and AI Blue; the defence of the defenders' own AI is that of AI Blue, Chapter 15; independence is that of AI Purple, Chapter 15.

What you will learn in this part:

• What a mandate manifest for an operations agent says
• How independent verification is specified
• How the control plane is protected
• How agents earn and lose levels

Part V places operations in the organisation.