Compile system outage notes into a post-mortem report, verifying root causes before distribution

• Gather incident logs, chat threads, and ticket updates from the support and engineering teams
• Reconstruct a chronological timeline of when the issue started, was detected, and was resolved
• Follow up with individual engineers via direct message to clarify conflicting notes on root causes
• Draft the summary report in a document template, synthesizing technical details into business impact
• Send the draft to department heads for review and revise based on their feedback before final distribution

Chasing down engineers to clarify vague troubleshooting notes while trying to build an accurate timeline under pressure.

1. You

   1. Assemble the incident record

   Export or copy the relevant monitoring alerts, chat threads, support tickets, deployment notes, and incident updates into one working set. Include timestamps, authors, and links or identifiers where available.

2. AI

   2. Build a sourced incident timeline

   Give the working set to AI and ask it to order events by time, separate detection from response and resolution, and attach each claim to its source. It should flag missing timestamps, duplicate events, and contradictory notes instead of filling gaps with assumptions.

3. AI

   3. Separate evidence from root-cause hypotheses

   Ask AI to group technical notes into observed symptoms, confirmed actions, possible causes, and unresolved conflicts. AI can identify which sources support each hypothesis, but it cannot independently verify a root cause from incomplete or conflicting notes.

4. You

   4. Confirm or reject the root cause

   An engineering or operations owner must decide whether the evidence supports a confirmed root cause, a provisional cause, or no conclusion, and resolve key conflicts with the responsible engineers. If the cause is not verified, mark it as unconfirmed and hold distribution or clearly disclose the uncertainty.

5. AI

   5. Draft the post-mortem

   Provide AI with the approved timeline, root-cause status, customer or business impact, response actions, and follow-up items. Ask it to produce the report in the required template while preserving source-backed facts and labeling open questions.

6. You

   6. Approve and distribute the report

   The accountable operations owner checks that the report matches the approved facts, removes unsupported claims, and decides whether it is ready for leadership distribution. Send the final report only after that decision; otherwise return it for correction or further investigation.

• mapping messy customer onboarding notes into our CRM and project management board
• Turn a recurring cross-team problem into a documented handoff
• finding out which team is dropping the ball on customer handoffs