Tech in Five - What are Agent Control Planes?

John Savill explains what an agent control plane is and why it matters when you start running AI agents at scale, focusing on the need for centralized oversight, evaluation, and cost governance.

Overview

The video introduces the idea of an agent control plane: a centralized layer for managing and governing AI agents, similar to how a traditional control plane provides coordination and oversight for a system.

Key themes covered include:

• What a control plane is

  • A conceptual refresher on control planes as the place where coordination, policy, and oversight live.

• Why AI agents need a control plane

  • As agents run across different environments and tools, there needs to be a consistent way to understand what they are doing and to keep them aligned with expectations.

• Analogy: air traffic control

  • Uses air traffic control as a mental model for coordinating many independent “actors” safely and predictably.

• Observe, evaluate, and improve

  • Emphasizes the need to:
    • Observe agent behavior
    • Evaluate outcomes
    • Improve agent performance and reliability over time

• FinOps considerations

  • Highlights cost governance as a core requirement when operating agents, tying agent operations to FinOps-style cost visibility and control.

• Build on what you trust

  • Stresses the importance of using trusted foundations (platforms, tooling, and governance) when deploying and operating agents.

Chapters

• 00:00 - Introduction
• 00:24 - What is a control plane
• 01:33 - Air Traffic Control
• 02:18 - Where are agents
• 03:01 - Control planes for agents
• 04:11 - Observe evaluate and improve
• 05:31 - FinOps
• 06:23 - Build on what you trust
• 07:29 - Summary
• 07:59 - Close