I don't understand the multi agent stuff people are building.
Mine is pretty simple:
1. Agent catches an error in production and kicks off an agent, which fixes the bug, runs the tests and opens a PR.
2. I review it over coffee.
This has freed up a lot of time for me to create more bugs. ☕
