Notes on operational work
Writing about coordination work, the design of assistants that propose rather than act, and what we're learning while building Zephyr. No launch announcements you didn't ask for.
The work that never gets scheduled
Every coordination job contains a second job that nobody planned, nobody measures, and everybody is quietly failing at.
The case for making the AI ask first
Every assistant product is being pushed toward autonomy. In operations work, that's the wrong direction, and the reason is asymmetry.
Waiting is half the job, and no tool tracks it
Task lists model work you will do. They have no representation for work you have asked someone else to do — which is most of coordination.
250 words, or it's a second inbox
A daily summary that takes five minutes to read is another thing to get behind on. The word limit is not a constraint on the feature — it is the feature.
Capture first, organise later — and let something else do the organising
The classic productivity advice is right about capture and wrong about who should do the processing.
An assistant that can say “I don’t have that recorded”
The hardest thing to build into a work assistant isn't knowledge. It's the discipline to return nothing.
Reading this because something keeps slipping?
That's the entire problem Zephyr is built for.