Prologue: Why a Second Book

The first book, The Engineering Philosophy of LLM Applications, was about how software engineering has to change when the core component of a system stops being a deterministic function and becomes a probabilistic language model. That book started from the model: the model is a probabilistic component that makes mistakes, and the engineering object is the application built on top of it.

This book starts somewhere else.

The first book was about how to live with an unreliable function. This one is about how to work with an unreliable colleague.

The two books touch in a few places: uncertainty, declarative prompting, type contracts, the portability of skills. The boundary is nailed down — anything that can be explained at the model layer is cited from the first book, not retold here. This book starts from the agent as an autonomous role and works its way up to teams and organizations.

Articles

• Not Another Agent Framework Guide — What this book is and isn't. It doesn't teach you how to use a framework; it asks how engineering should think when agents move into the core of a software system.
• A Personal Statement of Position — The author's technical preferences, core beliefs, and blind spots. Honest bias beats fake objectivity.