A Practitioner's Guide to AI Agent Infrastructure

The Invisible Stack

The Hidden Infrastructure of AI Agents

The orchestration layer, the execution layer, and the trust layer most teams forget to build.

Begin Reading

Chapters

  1. 01
    Why Now The confluence that made agents possible — and inevitable
  2. 02
    The Chain, the Graph, and the Loop Three metaphors for how agents think — and why each one broke
  3. 03
    The Tool Problem A protocol, a universal adapter, and the new attack surface it created
  4. 04
    Whose Harness? How every AI lab built the same thing differently — and what their choices reveal
  5. 05
    The Body Problem Sandboxes, runtimes, and the question of what agents are allowed to touch
  6. 06
    What Goes Wrong Failure patterns, antipatterns, and the lessons nobody writes blog posts about
  7. 07
    The Trust Boundary Identity, policy, and the question of who the agent is acting for
  8. 08
    The Personal Agent OpenClaw, and what happens when tens of thousands of people run agents without guardrails
  9. 09
    Skills and the Learning Loop The abstraction every team invented independently — and the agent that took it further
  10. 10
    The Third Layer Where the characters end up, what the book claims, and the questions the reader carries forward

Appendices

  1. A
    Timeline A chronological reference for the events, releases, and incidents referenced across the book
  2. B
    Architecture Diagrams Schematic representations of the major frameworks' internal designs
  3. C
    The Decision Framework A structured guide for evaluating and selecting agent frameworks and trust infrastructure
  4. D
    Glossary Key terms and concepts used throughout The Invisible Stack