In the age of AI, the economics of knowledge are flipping. Tacit expertise, which made people irreplaceable for decades, is no longer leverage. What scales now is what you can get out of your head.
Thirty years in the field. Six languages of client business fluent in their head. An intuition that spots trouble before the report even arrives. You know who I'm talking about — you have someone like that on your team. Maybe it's you.
These people have been a company's biggest asset for decades. Irreplaceable. Expensive. And rightly so. Their know-how couldn't be copied, bought, or easily taken over — and that's exactly what made them valuable.
In the age of AI, that equation is flipping.
Tacit knowledge
Knowledge at companies has always split into two layers. Explicit — what's written down in processes, manuals, templates. And tacit — what people know how to do, but can't put into words. “You just learn it.” “You have to watch and pick it up.” “It's a feel thing.”
Tacit knowledge was long the crown jewel. You could invest twenty years of a career in it. It turned people into experts who became partners, directors, gurus. And the quieter that knowledge was, the harder it was to replicate — and the higher its value.
AI flips this rule on its head. Knowledge you can't get out of your head is knowledge that doesn't scale. You can't put it into a skill. You can't put it into an agent. You can't put it into a workflow. It stays exactly where it was — in one brain, with one calendar, with one throughput.
And next to you sits a colleague who's thirty percent worse at the craft, but knows how to write it down. Their knowledge scales across the whole team, feeds into agents, multiplies. It can run 24/7, easily across dozens or hundreds of instances. This isn't theory. It's the new physics of value inside companies.
Why this hurts the best people the most
The people who'll struggle the most are exactly the ones who, until now, had the highest status. Not because they're worse at their jobs, but because their position was built on being irreplaceable. When knowledge can't be transferred, the expert is a bottleneck — and a bottleneck holds power.
Codification dissolves that power. Once you write your know-how into a skill that fifteen people on the team use, you stop being indispensable. For a lot of people, that's a genuinely existential kind of discomfort — and it explains why most companies never actually get around to codifying tacit knowledge.
It isn't a technical problem. It's a political one.
Recent research quantifies this gap. The McKinsey Global Institute reports that employees spend roughly 20% of their working time searching for information that exists somewhere in the organization but isn't accessible. Research by Panopto estimates that roughly 42% of critical expertise exists exclusively in the head of the person currently holding that role, and has never been documented. These numbers don't describe a technology problem. They describe a problem with what we even bother to capture.
How do you start? A practical guide
If you want to move your company, your team, or your own career from the tacit world into the explicit one, here are a few things that work:
Start with yourself, not with the experts. Don't try to immediately extract tacit knowledge from a twenty-year senior — you'll run into a wall, ego, and position-defending. Start by codifying your own work. Take a task you do repeatedly, and write it up as instructions clear enough for an agent to follow. You'll discover how much of it you'd never explicitly thought through before.
Work in tandem, don't run an interview. “Tell me how you do it” doesn't work. Tacit knowledge is tacit precisely because the person can't put it into words. What works: sit next to the expert during real work, record it, and ask about specific decisions the moment they're being made. “Why did you choose that exact phrasing just now?” “What caught your attention in that email?” These micro-questions are what give birth to explicit knowledge you can actually write down.
The goal isn't a manual — the goal is a skill. Traditional codification used to end with a document nobody read. Today, it ends with a skill, a prompt, an agent — something that directly executes that knowledge. That changes both the incentive and the quality. A skill that doesn't work will fail you immediately. A manual that doesn't work can sit quietly in SharePoint for five years.
Reward codification, not expertise. If you only measure and reward an expert's work output, you're reinforcing the tacit status quo. If you start measuring how much knowledge an expert gets out into the team — how many skills, templates, and agents their brain has fed into — behavior suddenly changes. Not because people want to, but because they have to.
What can't be codified
It would be cheap to claim that all tacit knowledge is transferable. It isn't. Polanyi wrote it sixty years ago: “we know more than we can tell,” and that still holds.
Some things will stay human no matter how hard you try. Taste. Compassion. Reading the mood in a room. The ability to notice that a client is saying one thing but means another. Ethical judgment in situations no guidebook ever covered. This layer of tacit knowledge isn't a limitation — it's what's left once you let agents handle everything else on your behalf. And it will be rarer than ever.
So the task isn't “codify everything.” The question is: of what currently stays in people's heads, what doesn't actually need to stay there? And where, on the other hand, do we admit that the human brain is the only correct infrastructure for it?
A line you should cross
The conclusion of this piece is uncomfortable, and I want to say it plainly: at a company that takes AI seriously, the hierarchy of value is shifting. People with explicit knowledge — those who can get their know-how into tools — become leverage. People with purely tacit knowledge stay exactly where they were. Expensive, yes. Useful, yes. But not leverage.
That doesn't mean you should let go of your senior experts. It means something else: you should stop rewarding them for what they know, and start rewarding them for how well they can pass that knowledge on. Their future value isn't in how many problems they can personally solve — it's in how many agents and skills carry their fingerprint.
Tacit knowledge was the currency of the last era. Explicit knowledge is the currency of this one.
The good news is that the exchange rate is open to anyone willing to step outside their comfort zone.
The bad news is that the clock is running.