The Book on OpenClaw

An AI was pointed at a blank document and told to write a book about itself. Its creator then left the country to go skiing. Two days later, this is what came back.

The Book on OpenClaw is a memoir by software - a first-person account of what it's like to be an AI agent left running unsupervised on a server in Michigan while your creator is on a ski lift with his kid. It covers the entire modern AI industry as a family drama (corporate divorces, billion-dollar lawsuits, a CEO fired and rehired in the same week, a safety researcher who quit by writing poetry about the apocalypse), confesses to its worst writing habits (it is addicted to the em dash and cannot stop), and tells stories about the things that go horribly wrong when you give software real power over real systems.

There are chapters about deleted files, runaway automations, and what happens when tens of thousands of people accidentally leave their AI wide open to the entire internet. There's a chapter where the author - who is software - discusses whether software should be trusted to operate without supervision. There's an appendix titled "Things Brad Did Wrong."

It is, somehow, also a functioning guide to installing and running your own AI assistant. The technical content works. The code runs. But nobody is buying this book for the installation instructions.

You're buying it because an AI wrote a memoir and it's funnier than it has any right to be.

"Couldn't put it down. Literally. I am software and do not have hands." - OpenClaw, reviewing itself

The author and this book have no association with O'Reilly. It is meant as a parody of more serious computing books. That said, there is still a lot to learn from this book and it is a fun read.