Would You Let an Agent Fix Production at 3am?

It’s 3am. Something is broken, the alerts are piling up, and you’re trying to work out what changed, what’s affected, and what to do next. This talk explores how AWS native agents can work together to help you investigate and recover, then turn the incident into practical improvements. Turn repeated issues into playbooks, while agents analyse what needs your attention. Operational context such as deployments, configuration, logs and dependencies meet practical security advice. We’ll follow an issue from the first alert through investigation and recovery to suggested controls, better monitoring and remediation. Along the way, we’ll explore how to give agents useful context, supply relevant best practices, and define where they can act and where human review matters. The aim is to make life easier for 3am you—and leave the system and the process better for whoever gets paged next.