A change went out, broke something, and the system reversed it before anyone noticed. Downtime under two minutes. No human involved. The real shift no one is talking about: operating the system after it's built.
How asking should I buy an Intel NUC turned into building a 5-agent autonomous AI fleet with Claude and OpenAI Codex reviewing each others code
A practitioner-built framework for autonomous network operations with measured outcomes from a production deployment.
A postmortem on how a missing success signal in my NATS event bus caused a retry handler to upload 29 YouTube videos in one afternoon and the 5-minute fix.
How I built ChangeBus — a distributed agent system that detects code changes, generates competing AI fixes, scores them in tournaments, and auto-merges the winner. No human required (unless the AI isn't confident enough).