video1.0<iframe src="https://www.loom.com/embed/230ee07a420a4fb7a5ecf5ff7f49d0f8" frameborder="0" width="1920" height="1440" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>14401920Loomhttps://www.loom.com14401920https://cdn.loom.com/sessions/thumbnails/230ee07a420a4fb7a5ecf5ff7f49d0f8-a8764e9e8c567d1d.gif591.616This Loom demos a prototype AI manager that aims to learn how employees work by analyzing Slack and GitHub signals rather than relying on internal documents. The author argues the biggest company asset is employees and claims the next AI organization will be the one that replicates effective human collaboration, with a key bottleneck being how agents work together. The system provides a “champion cockpit” for a manager, running simulated agent meetings to map each developer’s focus for the next day while humans continue doing the work. It visualizes an engineering pod’s activity as a live graph of recent commits and collaboration edges, tracks team heat and conflict, and includes settings like refresh time and Slack pings, plus an unapproved digest shield visible only to the champion. The prototype is built over about three weeks and is currently limited to developer signals and metadata, with code access not included yet.