Developer uses a 'chief of staff' AI agent to cap 20 concurrent sessions
Start here
Developer Kevin London says AI has made it too easy to start marginal work: on some days he runs as many as 20 concurrent agent sessions, only 5 of which matter, and marginal tasks expand faster than important ones. His fix is a "chief of staff" agent that reads a goals file, tracks three weekly priorities and pushes back on requests that fall outside them. Overnight agents are limited to labeled tasks such as fixing failing builds and merging pull requests. London presents this as his own weekly workflow, not a measured result.
Generated from reports · updated 57 min ago
Report timeline
Follow the reports to see the event from different sides.
- Hacker News · AIAI makes it too cheap to say yes
A developer describes running as many as 20 concurrent AI agent sessions, only 5 of which matter, and argues AI has made marginal work expand faster than important work. The fix is a "chief of staff" agent that reads a goals file, tracks three weekly priorities, and pushes back on off-goal requests, while overnight agents handle only labeled tasks such as fixing failing builds and merging PRs.
Heat for this event
Not enough continuous observations yet to draw a trend.