In 30 seconds

Much of the knowledge a workflow depends on lives only in experienced people's heads. To let software or new colleagues use it, make the relevant parts explicit, written down, agreed and owned, working with the person rather than extracting from them.

Why this comes up

An experienced employee "just knows" which customers need special handling, what a supplier's odd terms mean, or the four exceptions to a rule. This tacit knowledge is often the most valuable input to a workflow, and the easiest to overlook because it is invisible to the person who holds it.

You cannot automate a step that depends on knowledge nobody has written down. So making it explicit is groundwork, not paperwork.

Capture knowledge in the flow of real work

The best way to surface tacit knowledge is not a blank document titled "please write down everything you know." It is to work through real cases together.

  • Walk through recent examples and write down the decisions as they come up.
  • When someone says "I just know", stop and ask what tells them, and what would change their mind.
  • Capture their words first. Tidy the language later, after they confirm it is right.

The workflow discovery notes template gives you a structure for this.

Separate the kinds of knowledge

As you capture, sort what you find:

  • Stable rules: "Orders to this region always need a customs form." These can often become deterministic automation.
  • Facts: "This supplier's lead time is three weeks." These belong in a maintained source.
  • Judgment: "This wording usually means the customer is unhappy." This is interpretation, where AI may help, with review.

Also separate facts from instructions, because they age and are used differently. See how to separate facts from instructions.

Give every piece an owner

Knowledge that nobody maintains goes stale, whether a person or an AI uses it. As you write each rule or fact down, name who is responsible for keeping it correct. Ownership is what turns a one-time capture into knowledge that stays true.

A small example

A single dispatcher decided which shipments to hold in bad weather. Sitting with them through past decisions revealed a clear rule for two cases and genuine judgment for the rest. The rule was written down and owned by the logistics lead; the judgment cases were routed to a person. When the dispatcher later changed roles, the workflow survived.

What this does not solve

Making knowledge explicit does not keep it current, and it does not capture judgment that even the expert cannot articulate. Some knowledge stays with people, which is a valid reason to keep a human in the loop.

Good enough to move on when

  • The rules and facts a step depends on are written down.
  • They are recorded in the person's own words, then confirmed.
  • Each has a named owner.
  • You have separated stable rules, facts and judgment.