1 Incident Management AI Agent, Live in Production.

Postmortems are the highest-value artefact in incident management and the one most reliably not produced. The reason is scheduling, not neglect: the week after an outage is the week of catching up on everything the outage displaced, so the writeup is deferred to a quieter period that does not arrive. By the time it is attempted, the timeline has to be reconstructed from memory and chat scrollback, the contributing factors have collapsed into a single cause, and the action items are generic because the specific ones depended on detail nobody can now recall. The organisation then meets the same incident again and treats it as new.

Incident Postmortem turns rough incident notes into a blameless postmortem with a timeline, contributing factors and action items someone actually owns. Three things in that description are doing work. Blameless, because a writeup that identifies a person rather than a system guarantees the next incident is reported late. Contributing factors in the plural, because a single root cause is almost always a simplification that hides the two conditions which had to coexist. And action items with an owner, because an action item without one is a description of a problem. One agent covers this process; it drafts from the notes, and the analysis is reviewed by the people who were there.