All articles
AI + PM

The Book Is Out, and It Started in March

TL;DR

  • The book is out: The Project Manager’s Guide to AI, twelve chapters on building an AI workflow that holds up on a real project.
  • It runs on one case study from first prompt to day 90, artifacts shown as they went out, catches included. One of those catches is a citation an agent invented for a meeting that never happened.
  • The governing rule is four words: AI drafts; you judge. Dates, numbers, names, commitments, and tone get checked by a human, every time.
  • $19 as a PDF today, Kindle and paperback coming, and a $39 bundle with the Starter Kit if you want both halves.
  • I started writing it in March. The kit shipped a piece of the work early; the book is the rest of it.

Four days after the kit went out, the rest of the work ships. The Project Manager’s Guide to AI is a book, 196 pages, and I started it in March, which makes it older than the kit, older than the launch posts, and older than most of what this blog has said about AI since May. The kit was the piece that could go out early. The book is what a tool can’t carry: the method, the judgment, and the catches.

What’s in the Book

Twelve chapters in five stages: foundations, first wins, craft, systems and tools, then scale and guardrails, with a 90-day roadmap as the capstone. The whole thing runs on one case study carried start to finish, Project Aurora, a customer-portal launch with a $1.2 million budget, nine months, and a fourteen-person hybrid team.

The artifacts appear as they went out, margin-annotated. That’s the part I cared most about getting right, because a guide that only shows the polished version teaches the wrong skill. There are 27 copy-ready prompts, 12 Aurora exhibits, 45 figures, and appendices for the parts you’ll want to print.

The Rule, and the Catch

The book runs on a rule short enough to remember on a bad day: AI drafts; you judge. Five checks ride along on everything an agent touches: dates, numbers, names, commitments, tone. If the agent can’t show its work on one of those five, it doesn’t go out under your name.

The rule exists because of what happened in the case study. One of the agents invented a citation, for a meeting that never happened. That artifact went into the book instead of out of it. A case study where the AI behaves is a case study nobody will recognize on the day they need the book, so the failures are part of the record.

What the Kit Can’t Carry

The kit does the record-keeping. What it can’t carry is the reasoning underneath: why the checks exist, how to tell a summary you can trust from one you can’t, and how to stage agent permissions so the tools allowed to act are the ones you’ve vetted. The book covers that ground, which is why it came first in the writing order even though the kit shipped first in the calendar.

There’s a market number behind the timing, the one I cited when I announced this: most organizations are using AI somewhere, and very few have scaled it into how the work actually runs. The gap isn’t access anymore. It’s whether the people running the projects know which parts to hand over and which to keep. Closing that gap is a reading job before it’s a tooling job.

Out Today

The book is at thirtyyearpm.com/book: $19, PDF, instant download, 196 pages at 6×9. Kindle and paperback are planned; the page will say when. If you want the method and the tooling together, the bundle with the Starter Kit runs $39, which is $9 off buying them separately.

If you read two chapters before the rest, make them the pre-mortem and the whole-history review. They’re the two I highlighted when I announced it, and they’re the ones that change how the next project starts.

Shayne Yestal

Project manager, studio founder, and AI advocate with 30+ years of experience. Co-founder of Mocoda. Creator of ThirtyYearPM.