Knowledge extraction from departing employee

A skill triggered when an employee gives notice. Runs a structured interview protocol (50+ questions), cross-references their vault contributions and meeting transcripts, and outputs documented workflow runbooks and reusable agent skill files. The departing person's tacit knowledge becomes versioned artifacts before they walk out.

At 30–100-person teams, every senior departure usually means 6+ weeks of "wait, how did they do X?" Slack archaeology. Extracting their workflows into named skills before exit converts a knowledge loss into a permanent capability. The US has institutionalized this practice for senior departures; AI now makes it tractable in 2–4 hours instead of two weeks of off-boarding interviews.

1. Trigger when notice is given (day 0 of notice period).
2. Run the structured interview — 50+ questions covering: daily workflow, recurring rituals, tool stack with shortcuts, "who do you ask when X breaks", judgment calls you make weekly, the 3 things that would fall over if you left tomorrow.
3. Use voice + Wispr Flow for the interview — 2-hour session, transcribed live.
4. Agent reads the transcript + the person's vault contributions + their last 6 months of meeting transcripts.
5. Agent generates one runbook per recurring workflow + one skill file per agentic task.
6. Departing person reviews and corrects in their final week.
7. Successor (or interim owner) reads the runbooks day 1.

• Don't try to do this in the last 3 days. Start day 0 of notice — by week 3 the person is mentally already gone and the quality drops.