What this book is
A field guide written for engineers who already know how to ship — and are now figuring out how the job changes when an AI pair is in the loop on every commit.
It covers:
- A taxonomy of decision points in an AI-assisted workflow — where judgment lives, and where it moves.
- Replay-driven code review: defending design decisions when the reviewer wasn't in the chat.
- Failure modes of AI-generated code: silent regressions, plausible-but-wrong refactors, security drift.
- A practical methodology for debugging with AI that scales beyond toy bugs.
What it isn't
This is not a prompt cookbook. It will not teach you how to write better prompts. The book assumes you can already get the model to produce code. The question it answers is: now what?
Format
- ~340 pages
- Companion challenge set on the Codritium platform — every chapter has a real replay you can step through