In one sentence
Tacit knowledge is knowledge people hold through experience rather than in any document.
The simple model
Some of the most important company knowledge is never written down. An experienced employee may know that a certain customer needs a special approval, that one supplier uses unusual terminology, or that a "normal" request actually has four exceptions.
They rarely think of this as knowledge. To them it is just how the work is done. That is exactly what makes it easy to miss.
Why you care
Software and AI cannot use knowledge that only exists in someone's head. Before you can automate a step, the judgment behind it usually has to become explicit: written down, agreed and owned.
If you skip this, the automated version handles the easy cases and quietly fails the ones that depended on the unwritten knowledge.
Two kinds of hidden knowledge
It helps to tell two things apart.
- A stable rule the person has internalized: "Orders from this region always need a customs form." This can often become a deterministic rule.
- Judgment built from many examples: "This wording usually means the customer is unhappy." This is interpretation, and may be a place where AI helps, with review.
They need different treatment, so name which one you are looking at.
A small example
A dispatcher "just knows" which deliveries to hold when the weather is bad. Asked to explain, they describe a mix of route, cargo type and a supplier's reliability. Some of it is a clear rule. Some of it is a feel developed over years.
Making the rule explicit is straightforward. The feel needs either a documented set of examples or a human kept in the loop.
What this does not solve
Making knowledge explicit does not keep it current. Once written down, it needs an owner and a review habit, or it will drift out of date. See how to keep knowledge current.
