AI generated 156 items on his to-do list, and more work than it finished

The piece is a first-person commentary, and its starting point is a practical shock: a daily list that had grown to 156 items. The author is clear that he remains a big believer in AI and uses it to organize ideas and get thoughts onto a page. His complaint is narrower and more operational — an assistant can produce a plan, then a checklist, then subcommittees for the checklist, and each item can sound reasonable on its own.

What follows is a set of self-imposed rules aimed at the gap between activity and completion. He wants a daily list that reflects available capacity, related work grouped under one clear execution item, and proof attached to completion, such as a live link for something published or a delivery record for a file. He also wants blockers stated specifically and wants the system to count outcomes — how many pitches went out, which replies need an answer, what reached the client.

He acknowledges the changes require judgment, since some days have more real deadlines and some tasks take minutes while others need uninterrupted hours. Underlying it all is a simple test he now applies: what would "done" look like to the person who asked for this?