About
About the author
Author bio
I’m an AI product manager. Most of what I do is decide what gets built, what gets cut, and how the thing behaves when a real person is using it under pressure.
I’ve spent the last few years building with these tools daily, which is also how I ended up writing about them. The productivity gains are real and I’d defend them. What kept nagging was quieter: the reasoning I used to do myself was now happening somewhere I couldn’t see.
So I started reading the research instead of the marketing. Randomized studies, learning science, the work on cognitive offloading. Some of it confirmed what I suspected. A fair amount of it corrected me.
This book came out of that reading, and out of realizing my own habits had already shifted while I was busy being impressed with the output.
Photo
Why I wrote this
I wrote this because I watched my own judgment get quieter.
Not worse in any way I could point to. Just less exercised. I’d draft something, get a fast good-enough answer back, and ship it. The reasoning that used to happen in the gap between the question and the answer… it stopped happening, and nothing flagged it.
Then I started seeing it in other people’s work. Decisions that were defensible but not actually decided. Documents that read fine and fell apart the moment someone asked why.
The research points the same direction. A 2025 randomized study in the British Journal of Educational Technology found the ChatGPT group outperformed the others on the task itself, while knowledge gain and transfer showed no significant difference. Better output. Same understanding.
This book isn’t an argument for using AI less. I use it constantly. It’s a set of practices for keeping the part of the work only you can do… the comparing, the testing, the deciding under real constraints… in your own hands.
You won’t feel the difference most days. It shows up later, on the call you actually needed to get right.
How this book handles evidence
Three commitments run through the manuscript:
- 01
Famous statistics are checked against the studies that did or didn't produce them.
- 02
Contested science is named as contested, not smoothed into consensus.
- 03
The author's inferences are flagged as inferences in the notes, so the reader can see where the evidence ends and the argument begins.
Every named person in the book except the author is an invented composite, assembled from documented patterns and flagged as such on first appearance.