IncidentScribe for SRE & on-call teams
You ran the incident. Now there’s a 1,800-line Slack channel, a PagerDuty timeline, and a Confluence template with your name on it. The reconstruction is the part everyone dreads: scrolling back through the chaos, untangling who did what when, and turning it into something coherent while the details are still fresh enough to matter.
From channel to draft
Paste the export. IncidentScribe extracts a structured timeline — typed actors, timestamps, and summaries, each traced to its source line — then drafts the five sections every good postmortem needs: Summary, Timeline, Root Cause, Contributing Factors, Action Items. You edit, sign off, and export to Markdown or PDF. The walkthrough on a real memory-leak incident runs about three minutes wall time.
A draft you can actually defend
It’s not a one-shot summary you have to re-verify by hand. Every claim is two clicks from its source via the citation chain, and fabricated events never make it into the timeline. When the engineer who lived the incident reads the draft and says “that’s not quite right,” they click straight through to the log line and you settle it on the spot.
Works even when cloud AI doesn’t
Plenty of SRE teams are under a standing ban on pasting incident data into cloud LLMs. IncidentScribe runs Apple’s Foundation Models entirely on-device, so it works the same whether your org permits cloud AI or forbids it. There’s no shared workspace to get approved and no team-tier rollout — it’s a single-operator Mac app you can start using the same afternoon. (Teams that do allow cloud LLMs can opt into a Pro bring-your-own-key mode; it’s off by default.)