Field Notes · Published by WorkClaw
RSSWriting on autonomous agents, meeting follow-through, and what it takes to let AI act on real work, built to be read by people and answer engines alike.
Project management software solves memory and routing. Once agents do both automatically, by eliminating the input layer and holding the context that used to live in people's heads, the reason to open a tracker goes away.
Read dispatch →Most of the headcount a firm adds between 15 and 50 people goes to coordination, not delivery. Automate that layer instead of hiring for it, and a small team runs with the follow-through of a much larger one.
Read dispatch →Executives aren't skeptical about AI agents anymore; they want one. What stops them is a specific, rational fear about data and access. That fear comes down to governance, and governance is buildable.
Read dispatch →Work slips even when everyone is on top of it, because logging the work is a different job from doing it. The answer is an agent that lives where the work is already being discussed.
Read dispatch →The full workflow: how an autonomous agent reads a meeting transcript, extracts commitments, assigns the right owners, and drafts the follow-up in under five minutes, with a human approving before anything ships.
Read dispatch →