The sovereign-agent book
A hands-on book about building an AI agent organization that can be trusted to act on someone's behalf — not a diagram of one. Every chapter follows the same running example, a small ice-cream shop named Lucy's, from a single unreliable script to a governed system that recovers from its own failures and can be handed a real pilot.
Who it's for
Engineers and builders who already write code and want to put an AI agent in charge of real work — memory, money, or a schedule — without losing the ability to say what it did and why.
What you'll build
A working agent organization, piece by piece: a memory you can hold to account, a boundary around what the agent may touch, a recovery path when it fails mid-task, and a pilot it can run for a real business.
How to read it
The chapters build on each other in order — each one assumes the organization built in the ones before it. The table of contents below is the reading order.
- 0 Chapter 0 — Lucy's First Shift Not yet published
- 1 Chapter 1 — The organization remembers Not yet published
- 2 Chapter 2 — Work needs governance Not yet published
- 3 Chapter 3 — The Actor Is Not the Model Not yet published
- 4 Chapter 4 — Work stays inside its boundary Not yet published
- 5 Chapter 5 — Authority needs a fence Not yet published
- 6 Chapter 6 — The organization recovers Not yet published
- 7 Chapter 7 — The organization wakes itself Not yet published
- 8 Chapter 8 — The Store becomes a catalog Not yet published
- 9 Chapter 9 — Each product has its own threshold Not yet published
- 10 Chapter 10 — One signal wakes one need Not yet published
- 11 Chapter 11 — Replenishment scales without losing governance Not yet published
- 12 Chapter 12 — The pilot begins with a receipt Not yet published
Illustrated by Prof Rod
Written by Rod Rivera
The sovereign-agent book
Prof Rod
Cover treatment pending the operator's review at
/book/cover-proposals.
