capturing a system before it goes dark
A departing expert, a vehicle-data authoring service nobody else fully understood, and two Atlassian instances of scattered history — harvested, classified, and published as ten reference pages in a day.
lexus.com platform context; people, page ids, and ticket keys withheld. the service is described generically as a vehicle-data authoring service.
A vehicle-data authoring service feeding lexus.com — CMS authoring, a five-minute hash poll, the vehicle API, build orchestration, the build UI — was about to lose the one person who held it in their head. Rather than interview and summarize, I built a reproducible read-only harvester across two Atlassian instances: 674 wiki items and roughly 5,760 issues, run through a classifier that filters out the acronym collision with an unrelated VIN-related term (most of the raw hits), into 93 mirrored page bodies, inventories, a 2015→2026 change timeline, and a capstone assessment. An idempotent publisher then created ten reference pages under one wiki hub. The platform director's reaction — the unprompted kind — was that it was the best documentation of the system anyone had seen, and it became the basis for an IT approval that had been stuck.
The org decision followed from the artifact: maintenance mode through the quarter, Slack intake routed to a ticket label, biweekly review, named owners. Knowledge capture as a pipeline, not a meeting.
Full case study in progress — artifacts pending clearance or writing time. The insight below is already earned.
Institutional knowledge is a system with a bus factor of one. Treat its capture like a data pipeline — harvest, classify, mirror, synthesize, publish — and the output is reproducible instead of one more interview nobody wrote up.
the insight