The Robot Playbook · A founder's story
Build it. Then build it properly.
Two books on building real software with AI agents, from your first site to a way of working you can trust.
The Robot Playbook is a three-book series. Books 1 and 2 are out. Ship It is on the way.
The Robot Playbook · Book 1
Start Here:
It Works on
My Machine
Your first site with an AI agent, from the idea to live
A field workbook · Early Edition · A founder's story
Mark Wetherell
The Robot Playbook · Book 2
Don't Let the Robot Push to Main
Agile, rebuilt for a team of AI agents
A field workbook · Early Edition · A founder's story
Mark Wetherell
Three rules from Book 2
Rule 4
The prompt is a user story.
Who it's for, what done looks like, and the fences an AI teammate can't guess.
Rule 7
A check that passes without having run is worse than no check.
Every test fails once, on purpose, before it's trusted.
Rule 9
The agent merges. The human pushes.
Nothing reaches real users that a person didn't send.
This site is the demo
Every page here was built the way the books say.
I built PocketSense with this method, then built wetherell.org to test it again from scratch. The build log shows the real lanes, the real bugs, and what each rule caught.
Read the build log
The app I always meant to build
I started in accounting, then spent a career building and testing software, much of it on Scrum teams. My family lived the paycheck juggle for years.
PocketSense is the app I always meant to build. I built it with AI agents, and this site is where I write down how.