Philosophy to Operations

Michael Polanyi for AI builders and agent designers

Polanyi matters because modern AI repeatedly runs into the same wall: explicit instructions are not the whole game. Experts rely on tacit knowing.

We know more than we can tell

This line is not anti-knowledge. It is anti-reductionism. It means that some of the most important parts of performance are integrated in practice, attention, and embodied judgment before they are verbalized.

Focal and subsidiary awareness

Experts do not just know more facts. They know which details to rely on in the background while staying oriented toward the whole task. That is why attention switching is central to strong diagnosis and strong review judgment.

Indwelling and skill acquisition

People do not learn tacit skills through explanation alone. They learn by dwelling in practice. That is why real transfer needs drills, simulations, apprenticeship, and new-case validation rather than lecture-only instruction.

Why this matters for agent design

If you ignore Polanyi, you will try to build agents from explicit rules only. If you take Polanyi seriously, you will spend more time extracting cues, failure modes, escalation boundaries, and transfer tests. That changes the quality of the system you build.