There is a quiet assumption underneath almost every AI-for-coaches tool on the market right now. It goes like this: your expertise is contained in your content. Feed the machine your books, your courses, your recorded calls and your written frameworks, and it will have you. It will know what you know.
It will not. And there is now academic research that explains precisely why.
The assumption fails because the relationship between what an expert has written down and what an expert actually knows is far weaker than the tools depend on. Recent work on expertise and knowledge capture suggests that standard content ingestion captures something in the region of twenty percent of what an experienced practitioner genuinely knows. The other eighty percent never made it onto the page, because it was never the kind of thing that goes onto a page in the first place.
If you build an AI version of your practice on the twenty percent, you get a competent summary of your published material. What you do not get is you.
What tacit knowledge actually is
The gap has a name in the research literature. It is the distinction between explicit knowledge and tacit knowledge, and understanding it is the difference between an AI tool that sounds like your newsletter and one that sounds like your practice.
Explicit knowledge is everything you can articulate and write down. Your models, your frameworks, your core concepts, the material in your book. It is real and it matters, but it is the surface. It is the part of your expertise that was always going to be communicable, which is exactly why you already communicated it.
Tacit knowledge is everything you know how to do but have never fully put into words. It is the judgement that tells you which of your frameworks this particular client actually needs, and which would be a mistake to reach for right now. It is the thing you notice in the first ninety seconds of a session that changes your entire approach before the client has finished their opening sentence. It is knowing when to stay silent, when to push, when the presenting problem is not the real problem, and when to abandon your own plan because something more important just walked into the room.
None of that is in your book. Not because you held it back, but because it is not the kind of knowledge that survives being written down. It lives in the doing, in the thousands of hours of practice that built a set of instincts you can deploy but cannot fully explain. That is the eighty percent. And it is the entire reason clients come to you rather than reading a summary of your method.
Why the whole market is building on the wrong fifth
Here is what makes this more than an interesting distinction. Almost every tool in the AI-for-coaches space is built on content ingestion. Upload your material, and the tool learns you from it. The entire category has organised itself around the twenty percent, because the twenty percent is the part that is easy to capture. It already exists as files. You can drag it into a box.
The eighty percent cannot be dragged into a box, so the market has quietly agreed to pretend it is not the important part. Tools compete on how much content they can ingest, how well they mimic your written tone, how convincingly they reproduce your explicit frameworks. They are competing over the surface, and calling it depth.
This is why so many practitioners have the same experience when they build an AI version of themselves. They test it, and it is fine, and fine is precisely the problem. It sounds like their material. It does not think like their practice. It answers the way their FAQ page would answer, not the way they would answer in the room, because the thing that makes them answer differently in the room was never captured. It was never even sought.
The tool is not broken. It is doing exactly what it was designed to do with exactly the inputs it was given. The inputs were the wrong fifth.
What it takes to reach the other eighty percent
Tacit knowledge does not surrender itself to a content upload, but that does not mean it is unreachable. It means it has to be reached deliberately, through a different kind of process.
You do not capture tacit knowledge by asking someone to submit what they have already written. You capture it by asking the questions that make the invisible reasoning visible. Why did you make that particular move with that client. What did you notice that told you to change direction. What do you do when the framework does not fit. What would you never do, even though the textbook says you could. How do you know when to stop.
These are not questions a practitioner has usually answered out loud, because nobody has asked them in the right way, and because the knowledge is so automatic to the expert that they often do not realise they possess it until it is drawn out. It feels like common sense to them. It is not. It is twenty years of pattern recognition compressed into an instinct, and it can be surfaced, but only through structured extraction rather than passive ingestion.
This is the difference between building an AI tool from your filing cabinet and building one from your actual expertise. One is a retrieval system for things you already said. The other is an attempt to capture how you think, which is the only thing worth capturing.
The question worth asking before you build
If you are considering building an AI version of any part of your practice, there is one question worth sitting with before you start, and it is not a technical one.
What do I know that I have never written down.
If the honest answer is not much, then content ingestion will serve you perfectly well, and you should use it. But if you are an experienced practitioner, the honest answer is almost certainly a great deal, most of it the most valuable knowledge you have, and none of it sitting in the material a tool would ask you to upload.
The practitioners who understand this will build something categorically different from those who do not. Not a better summary of their published work, but a genuine extension of their judgement. The tools that only ingest content will keep producing competent reproductions of the twenty percent, and calling it a clone. The real opportunity, the one almost nobody is building toward yet, is in the eighty percent that was never on the page. That is where your actual expertise lives. It is also, not coincidentally, the part that no one else can copy.
