This page is the guide: three minutes to decide whether the book deserves your time.
What the four parts say, who they are for, a 12-point deployment checklist. The full text is in the PDF, free.
20 public, verifiable deployments: the workload of 700 support agents absorbed, QA headcount halved, one mine earning millions more a year. Every case detail now lives on the Evidence page, graded by source — no download needed for that.
Never the model. Four causes: not wired into core systems, no way to count the benefit, data that cannot leave the premises, no acceptance discipline. Each dissected, with real cost comparisons for build vs. buy vs. co-build. This part exists only in the PDF.
Four sentences: define "good" before you start; nothing goes live below the threshold; every action leaves a trace; data stays in your domain while models come to it. The core is distilled into the public acceptance baseline standard — usable without ever talking to us.
With a 2–4 week Bootcamp. By day 90 there is a running system on your desk and a set of numbers that adds up. The steps, the pricing, even "when not to choose us" — all spelled out on the Bootcamp page.
Parts 1 and 4 only: what others saved, and where you would start. About ten minutes.
The "nobody can count the benefit" failure mode in part 2, and the audit method in part 3: how the budget survives, and what you can sign.
Part 3 in full: acceptance baselines, gateway governance, deployments where data never leaves your domain.
If three of the twelve have no answer, the book is worth your time — or skip it and book a Bootcamp.