In 1966, the philosopher Michael Polanyi drew a line that nobody in business paid much attention to. On one side, explicit knowledge — the things you can write down, codify, put in a manual. On the other side, tacit knowledge — the things you know how to do but cannot easily explain. "We can know more than we can tell," he wrote.
For sixty years, this was a footnote in management textbooks. An interesting distinction. Occasionally referenced in a knowledge management pitch deck. Never urgent.
AI just made it urgent.
Explicit knowledge is what AI already has
Every regulation ever published. Every textbook. Every guidance note, every public filing, every professional standard. AI was trained on all of it. When you ask Claude or ChatGPT about MLR 2017, the SRA Code of Conduct, or FCA SYSC requirements, it already knows. It does not need your help with explicit knowledge.
Neither does your competitor's AI.
This is the part most firms have not yet understood. If you and your competitor both ask AI to draft a client engagement letter, you will get substantially the same output. The same structure. The same clauses. The same tone. Because you are both drawing from the same pool of explicit, public, universally available knowledge.
That is not a competitive advantage. That is commodity.
Tacit knowledge is what AI cannot get without you
Here is what AI does not know:
That certain clients need chasing on Monday mornings because they go quiet before they churn. That the FCA tends to ask about X before it asks about Y. That the wording in clause 7 of your standard terms matters more than the wording in clause 12 — even though clause 12 is three times longer. That when a client says "we are fine" they almost always mean "we have not looked yet."
None of this is written down anywhere. It lives in the heads of the people who do the work. AI cannot scrape it from the internet. Cannot train on it. Cannot infer it from public data. The only way tacit knowledge gets into AI is if someone deliberately puts it there.
This is the knowledge that makes your firm your firm. Not the policies. Not the templates. The judgment. The pattern recognition. The hard-won understanding of what actually works, developed over years of doing the work.
The paradox nobody talks about
Here is the uncomfortable truth: the easier knowledge is to capture, the less valuable it becomes.
Explicit knowledge is easy to document and easy to replicate. Which is exactly why AI commoditised it overnight. The moment a large language model can reproduce the contents of every published regulation and every standard template, the commercial value of holding that knowledge collapses. Not to zero — having it structured and accessible still matters — but to parity. Everyone has it. Nobody gains from it.
Tacit knowledge is the opposite. It is difficult to capture. Difficult to transfer. Difficult to replicate. Which is exactly why it is now the only sustainable source of competitive advantage. The firms that find a way to capture their tacit knowledge and make it available to AI will produce outputs that no competitor can match. The firms that do not will run the same AI as everyone else and wonder why everything they produce sounds generic.
You would recognise it if you saw it
The senior partner who can look at a prospect and know within five minutes whether they will convert or waste three months of your time. That is tacit knowledge.
The compliance officer who reads a suspicious activity report and immediately sees what is missing — not because there is a checklist, but because she has read a thousand of them. Tacit knowledge.
The operations manager who knows the real deadline is two days before the stated deadline, because the regulator portal crashes every time and you need a buffer. Tacit knowledge.
Now compare: your AML policy document. Explicit. The FCA handbook. Explicit. Your engagement letter template. Explicit.
AI already has a version of all the explicit knowledge. It needs the tacit. And it can only get the tacit from your people.
Most firms get this backwards
The first instinct when firms adopt AI is to feed it documents. Upload the policy manual. Import the templates. Attach the procedures guide. This feels productive. It feels like you are "training" the AI on your firm.
You are not. You are giving it explicit knowledge — the kind it already has a version of. It is not worthless, but it is not what differentiates you.
The firms that pull ahead are the ones who capture the other kind: the judgment calls, the client intelligence, the operational nuances, the "we tried that and it does not work" institutional memory that exists nowhere except in heads.
That is harder. It requires asking people questions they have never been asked. It requires making it easy for them to contribute what they know in a way that does not feel like writing a manual. But it is the only path to AI that sounds like your firm instead of every firm.
Tacit knowledge has a shelf life
People leave. People retire. People forget. Every week, tacit knowledge leaks out of your firm and nobody notices because it was never written down in the first place.
The senior associate who has been with you for fifteen years carries an extraordinary amount of institutional knowledge. When she leaves, that knowledge leaves with her. The new hire starts from zero. Every client relationship, every operational shortcut, every hard-learned lesson — gone.
AI does not solve this problem. But a system that captures tacit knowledge and makes it available to AI solves it permanently. Every insight captured today makes every AI interaction better tomorrow. And the day after that. And the year after that. It compounds.
The platform exists to bridge the gap
A constitution captures how your firm thinks — your voice, your values, your standards, your guardrails. Knowledge hubs capture what your firm knows — the domain expertise, the client intelligence, the operational reality that makes you different.
Together, they are the mechanism that turns tacit knowledge into structured intelligence that AI can use. No manual writing. No database building. Answer questions. Contribute insights. The platform does the rest.
Polanyi was right. You can know more than you can tell. But now, for the first time, there is a way to tell it — and a reason that matters more than ever.