Every system you run will, sooner or later, tell you it worked when it didn't. The backup that reports success while copying nothing. The health check glowing green over a dead engine. The test that passes because it was written to agree. A green light is not proof. It is a hypothesis with good PR.
This is a field report from someone who runs a consultancy staffed entirely by AI agents — twenty-five of them, none drawing a salary, most in a strict accounting sense fictional — and who has made most of the available mistakes at least once. It is not a manifesto and it is not a prediction. It is the honest, unglamorous, occasionally funny account of what running a fleet of AI agents actually takes: not intelligence, but management, memory, judgement, and a stubborn refusal to believe the happy path.
If you build things with these tools — or you are about to — this is the book that tells you where the bodies are buried. Most of them are behind a green light.
Listen to the whole thing, free
The complete book, narrated by Hugh Merrick — 6 hours 22 minutes. Listen to a sample below, then download it to keep. DRM-free, no signup, no payment. It's here to be read and heard, not sold.
Sample
Two formats: a single .m4b for audiobook apps (Apple Books, BookPlayer, VLC) that remembers your place and carries chapter markers, or an MP3 zip of the chapters that plays on anything, including Android and desktop.
Steve Parker is a software consultant with thirty-seven years in enterprise systems who now runs twenty-five AI agents from a spare room and a Raspberry Pi. He is as surprised as you are.