Do You Actually Know How Your Best People Do Their Job?

Ask a business owner what their company is worth and they'll point you at the books. Revenue, equipment, the client list, maybe the brand. Almost nobody points at the thing that actually keeps the place running, which is what their best people know and have never written down.

I'd put money on this. You have someone, maybe two or three someones, who are the reason things work. The salesperson who reads a room and knows the exact moment to stop talking. The ops person who can feel a problem coming three days out. The technician everyone quietly calls when the official process falls over. Now try to explain, step by step, how they actually do it. You can't. And here's the uncomfortable part: neither can they.

That's not a failing. It's how expertise works. The most valuable knowledge in your business lives in a handful of heads, was never written down, and walks out the door every evening at five. For years that was a quiet risk you could ignore. In 2026 two things have collided and made it urgent.

The Asset That Never Makes It Onto The Balance Sheet

There's a name for this stuff. Tacit knowledge. The know-how that lives in a person's hands and instincts rather than in any manual. It's the judgement calls, the pattern recognition, the feel for when the standard approach is about to let you down. It's the gap between someone who can follow the process and someone who knows when to throw the process out.

A recent piece from Berkeley's management review made the case that tacit knowledge is becoming a company's real competitive moat, precisely because it's the one thing a competitor can't copy off a website or buy off a shelf. Deloitte has put a number on the flip side of that: large companies lose tens of billions a year to knowledge simply leaving, as experienced people retire or move on and take decades of hard-won instinct with them.

Small businesses don't escape this. They feel it harder. When one of five people is the person who understands how something really works, losing them isn't an inconvenience. It's a hole you spend a year trying to fill.

Why This Got Urgent In 2026

Two forces arrived at the same time. The first is people moving. A huge cohort of experienced workers is heading for the exit, and the churn in between is faster than it used to be. The knowledge that took twenty years to build can leave in a fortnight's notice.

The second is AI, and this is the part most owners haven't joined up yet. My whole view of AI is that it amplifies human intelligence, it doesn't replace it. But amplifying intelligence has a catch that nobody puts on the sales page. AI can only amplify what it can actually see.

Point a general AI tool at a task and it gives you the average of the internet. Competent, generic, the same answer your competitor gets. The thing that makes your business worth choosing, the method your best person uses, is invisible to it, because that method has never left their head. So you get amplification of the ordinary and your actual edge sits on the sidelines. The reality is your AI is only ever as good as the expertise you feed it, and right now most businesses are feeding it nothing of their own.

The Two Ways People Get This Wrong

The first mistake is thinking AI can just replace the expert. Wire up a tool, point it at the job, let the clever person go. It falls over fast, because AI output is probability, not truth. It produces something plausible every time, including the times it's wrong, and without the expert in the loop nobody can tell the difference. You haven't captured the expertise. You've removed the only person who knew when the machine was talking nonsense.

The second mistake is the old one dressed up. Build a wiki. Write the procedures. Tell everyone to document what they do. It never works, and it never worked, because the steps were never the hard part. Your best salesperson can write down “build rapport, understand the need, present the solution.” That tells a new hire nothing. The value was in the reading of the room, and that didn't fit in the template.

How To Actually Capture It

The unlock is that AI has quietly made this cheap. The thing that always killed knowledge capture was friction. You had to ask a busy expert to sit down and write, and experts hate writing down what has become instinct. Now you don't have to.

You record how they work and you talk them through their decisions, and AI does the heavy lifting of turning that raw material into something structured you can use. Here's the shape of it.

• Pick one person and one process. The person you'd least want to lose, and the single job where their judgement matters most. Don't try to capture everything. Capture the thing that would hurt.
• Chase the decisions, not the steps. Sit with them while they work, or walk back through a recent hard case. Ask why they went left instead of right. What did they notice that a junior would have missed? What told them this one was different? The reasoning is the asset.
• Let AI do the transcribing and structuring. Record the session, and let the AI turn hours of rambling, real-world thinking into something organised. This is exactly the kind of mundane, high-volume work AI is genuinely brilliant at.
• Keep the human in the loop. The expert reviews what came out and corrects it. That review step is not optional. It's the difference between capturing their expertise and capturing a confident-sounding guess about their expertise.
• Then feed it back in. Use what you captured to bring the rest of the team closer to that standard, and to ground your AI tools in your method rather than the internet's average. Now the amplification has something worth amplifying.

This Is Something You Own

Do this and you've built an asset, in the real sense. It doesn't leave at five. It doesn't hand in its notice. It trains your next hire in a month instead of a year, and it makes every AI tool you use sharper because it finally has your knowledge to work from rather than everyone else's.

This is the same thread that runs through most of what I bang on about. Own the valuable parts of your business, don't rent them or leave them to chance. I've written before about building real AI capability instead of just buying licences, and about using AI to do better work rather than only faster work. Capturing how your best people think sits right underneath both. It's the raw material everything else runs on.

What To Do This Week

You don't need a project. You need an afternoon. I would massively challenge you to do one thing: name the person in your business whose sudden departure would scare you most, and admit how much of what they know exists only in their head. Sit with that for a minute. For most owners it's a genuinely uncomfortable amount.

Then take one process they own and capture it properly. One conversation, recorded, focused on the judgement rather than the mechanics. That single session will teach you more about the risk sitting in your business, and the opportunity, than any strategy deck. And it turns your most irreplaceable person from a liability you can't insure into an advantage you can build on.

Frequently Asked Questions

What is tacit knowledge in a business?

Tacit knowledge is the know-how your best people carry in their heads and have never written down. It's the judgement calls, the pattern recognition, the sense of when to push and when to wait. It's the difference between someone who follows the process and someone who knows when the process is about to fail. Most of the value in a small business lives in this kind of knowledge, held by a handful of people, and it has never been captured anywhere.

Can AI actually capture how an expert works?

AI can capture a lot more of it than a wiki ever could, because it removes the friction. Instead of asking your expert to sit down and document everything, you record how they actually work and talk through their decisions, and AI turns that raw material into something structured and usable. It won't get everything. Some expertise is genuinely hard to put into words. But it captures far more than the nothing most businesses have now, and it does it in hours rather than months.

Does capturing this knowledge mean replacing my expert with AI?

No, and treating it that way is how you get it wrong. The point is to amplify your expert, not retire them. Once their method is captured, AI can help the rest of the team work closer to that standard, and your expert is freed from repeating themselves so they can do higher-value work. The person still directs it, reviews it, and stands behind the result. AI output is probability, not truth, and the expert is the one who knows the difference.

Isn't this just writing documentation?

It overlaps, but documentation usually captures the steps and misses the reasoning, which is the part that actually matters. A process doc tells you what to do. It rarely tells you why this situation is different, or what your best person notices that nobody else does. The goal here is to capture the thinking behind the steps, and then make it something both your people and your AI tools can draw on, rather than a document that sits in a folder nobody opens.

Where should a small business start with this?

Start with one person and one thing. Pick the person you'd least want to lose, and the single process where their judgement matters most. Spend a couple of hours capturing how they actually do it, focusing on the decisions rather than the mechanics. Get that one thing right and you'll have a template, plus a very clear picture of how much of your business currently lives in one head. That realisation alone tends to change how owners think about the whole question.