Transformation

Tacit Knowledge

Tacit knowledge is the expertise, judgment and understanding that experienced people carry but have never formally documented, making it invisible to AI systems until it is deliberately externalised into a form they can access.
DATE UPDATED: 11 Sep 2026
|
FIRST PUBLISHED: 22 Mar 2026

What tacit knowledge is

The philosopher Michael Polanyi described tacit knowledge as the things we know but cannot fully tell. A senior lawyer who reads a contract and immediately identifies the clause that will cause problems in three years is drawing on tacit knowledge. They cannot necessarily explain every step of the reasoning that produced that judgment. They have simply read enough contracts, seen enough disputes and developed enough pattern recognition that the insight arrives faster than the explanation for it.

Organisations are full of tacit knowledge. It exists in the judgment calls that experienced people make routinely and that junior people learn by watching them. It exists in the informal processes that developed because the formal ones did not quite work. It exists in the knowledge of which clients need to be managed a particular way, which internal approvals take longer than expected and why, and which technical approaches have been tried before and failed. None of this is documented. Much of it is not even consciously held. And almost all of it is invisible to AI.

Why tacit knowledge matters for AI programmes

An AI system can only work with what it has access to. If the most valuable knowledge in an organisation exists exclusively in the heads of experienced people, the AI system built on that organisation’s data will not have access to it. It will produce outputs based on the explicit, documented information it can retrieve, which in a knowledge-intensive professional services firm is often a fraction of what actually determines quality outcomes.

This is why many AI programmes in professional services produce systems that work well on structured, routine tasks and poorly on anything that requires the kind of judgment that distinguishes a senior practitioner from a junior one. The structured tasks are documented. The judgment is not.

Tacit knowledge extraction is the process of making that judgment visible and accessible. It involves structured conversations with experienced practitioners, analysis of the decisions they make and how they make them, and the translation of those patterns into forms that can inform AI system design, whether as training data, retrieval content, decision rules or human-in-the-loop criteria that reflect where experienced judgment is genuinely required.

Tacit knowledge and the Audit stage

The Audit stage of the Arquin Framework pays specific attention to where tacit knowledge lives in an organisation, because this determines both the ceiling on what AI can reliably do and the most significant opportunity for value creation. A process that depends heavily on tacit knowledge is not necessarily unsuitable for AI. It is a process where the design work of externalising that knowledge is a prerequisite for building an AI system that is genuinely useful rather than superficially functional.

Insight

Professional services organisations are unusually dependent on tacit knowledge because their product is judgment. The expertise that differentiates a strong firm from an adequate one is rarely written down. It is accumulated through experience, transmitted through proximity and expressed in the quality of decisions that experienced practitioners make. This creates both the biggest challenge and the biggest opportunity in professional services AI programmes. The challenge is that the most valuable knowledge is the hardest to make available to AI. The opportunity is that the firms which successfully externalise it will have built a knowledge infrastructure that compounds in value in ways their competitors, whose knowledge remains locked in individuals, cannot easily replicate.

Misconceptions

Tacit knowledge cannot simply be captured by asking people to document what they know. Much of it is genuinely unconscious. Effective tacit knowledge extraction requires structured observation, pattern analysis and iterative refinement rather than a documentation exercise.

Tacit knowledge extraction is not a one-time project. As experienced people leave and new ones develop expertise, the tacit knowledge base of an organisation changes. The infrastructure for capturing and maintaining it needs to be ongoing rather than periodic.

Not all tacit knowledge needs to be externalised for AI to be useful. The goal is to identify the tacit knowledge that is most consequential to the quality of the processes AI will support, and to focus extraction efforts there rather than attempting to document everything.

BACK TO GLOSSARY