Shelf Fundamentals
BAA 101
Building AI Agents 101
Building AI Agents 101
Living book· 11 chapters
Summary
The basics of building an AI agent, and only the basics. Each chapter comes from a lesson learned on a real project, then is stripped of its specifics so it holds for any agent.
Contents
open-ended- 01Verify the effect, not the claim
- 02The agent loop and its exits
- 03The tool as the model sees it
- 04Data is never an order
- 05The model's output is untrusted data
- 06Refusal is a state
- 07The right page must arrive
- 08Read the tool call, not the answer
- 09Evaluate without lying to yourself
- 10Memory has an owner
- 11What must always happen is enforced in code
- Chapter 12