On what horizon to judge a system
A year spent moving from one client to another, seeing up close how an AI budget line is approved or rejected in committee, and over what horizon a system should be judged.
A year spent moving between clients, from an industrial group’s digital factory to a pharmaceutical company’s accelerator. Changing client every six months teaches that a system is judged by what it becomes after the people who built it have left. Next year should bring longer engagements, which is what I am looking for.

AI is treated as an investment to be justified, with criteria for stopping set in advance. The most useful chapter explains how to end a project without presenting it as a success.

It proposes judging a system by what it becomes when nobody looks after it any more. That is the test of a handover, and I have not found a better one since.

A century of thinking about complexity, in few pages. The main point for a practitioner: wanting to simplify a system almost always means ignoring part of it.