Photo — sticky notes on a monitor
Undocumented exceptions
The steps nobody wrote down because “everyone knows”. Until the one who knows is on leave.
/ The problem
When one person is out, the process is out.
How a process actually works, why a decision was made, which exceptions matter — most of it is never written down. Asking people to document everything creates work they won’t finish. A one-off interview can’t keep anything current.
Photo — empty desk, box of files
Handovers that don’t hold
Two weeks of notes, a shared drive, then the questions start the day after they leave.
Photo — expert surrounded by colleagues
Experts as bottlenecks
Interrupted ten times a day for what only they know. Nothing moves when they’re out.
/ How it works Step 01 Read the full process →
Interview. Structure. Find the gap. Interview again.
Photo — scoping session, whiteboard
Photo — employee on a call, headset
Screenshot — a wiki page with sources
Photo — calendar / follow-up scheduled
01
Identify what matters
Critical workflows, known dependencies, a handful of experts. Agree scope and an interview time budget. No company-wide mapping first.
02
Interview employees
AI agents run voice or chat interviews about real work: concrete examples, decisions, exceptions, failure cases, reasoning. Interviews surface who else holds knowledge.
03
Build the living knowledge base
An interlinked, searchable wiki. Every important claim keeps its source, owner, review status, and verification date.
04
Find gaps, schedule the next interview
Missing steps, contradictions, unresolved dependencies become focused questions for the right person — proposed and scheduled within their availability.
04 loops back to 02. That loop is the product.
/ What you receive
Three things you can act on.
Delivered into the tools you already use. Every claim keeps its source, owner, and verification date.
Living, source-backed knowledge base
Searchable and usable for daily work, onboarding, and handovers.
Dependency and coverage report
Which critical workflows still lack documentation or a tested backup — and who they depend on.
Prioritized automation backlog
Repetitive tasks with documented inputs, rules, exceptions, and approvals — validated before anything is automated.
/ Who it’s for
Operations and technical-services teams with a continuity problem they can name.
A key expert is leaving, retiring, or going on extended leave.
A concrete handover is coming and the existing materials aren’t enough.
A few people are interrupted constantly because only they know how something works.
You want to automate repetitive work but nobody can fully describe it yet.
5–10 critical processes per pilot
~5 experts, one team
1 test that matters: can a backup do it alone?
/ Questions
All questions → Before you ask.
Isn’t this just documentation?
Documentation is the by-product. The product is the loop: each interview exposes gaps, and the system brings the right person into the next conversation. Success is a backup completing the task without the expert.
How much of my team’s time does it take?
You set the interview time budget. Questions are batched, never repeated, and follow-ups are scheduled within approved availability.
Who owns the knowledge?
You do. Participation and recording are transparent, access permissions are enforced, and you control review and deletion.
Start with one team and the processes that would hurt most.
A fixed-scope paid pilot. We measure whether a backup can do the job before and after.