You do not have an information problem. Nobody does anymore. Whatever you're stuck on right now, the information you'd theoretically need is a search away — has been for twenty years, and an AI has made it faster again. The bottleneck was never access to facts.
The bottleneck is knowing what to do with them.
That's the thing I want to pull apart in this piece, because most of how we use AI today is still aimed squarely at the wrong target. We ask it to explain. And it explains beautifully. But explanation is the part you didn't need help with. What you needed was judgment — someone who's done this a hundred times looking at your specific situation and telling you the move. That's not information. That's expertise. And for the first time, you can download it.
We've been getting expertise the same way for centuries
Think about how you've actually acquired expert help across your life.
For a few hundred years, the answer was books. The expert writes down what they know, you read it, and you do the considerable work of translating their general principles into your specific mess. Books are wonderful. They are also entirely on you to apply.
Then came courses — books with video and a progress bar. Slightly more engaging, same fundamental deal: the expert teaches, you translate. We all know how that goes. The course completion graveyard is full of things we genuinely meant to apply.
And at the top end, there are consultants and coaches. This is the good stuff — an actual expert looking at your actual situation. The reason it works is that they do the translation for you. They don't hand you a framework and wish you luck. They apply their judgment to your problem and tell you what they see. The catch is obvious: it costs $500 an hour, there's a two-week wait, and most people never get near it.
So the pattern across the whole history of expertise is a trade-off. Cheap and scalable but you do the work (books, courses). Or someone does the work for you but it's expensive and it doesn't scale (consultants). You could have reach or you could have application. Never both.
A skill is the first thing that gives you both.
What "downloading expertise" actually means
A skill is an expert's methodology — their real decision-making process — packaged so an AI can apply it to your work. Not explain it. Apply it. The distinction is the entire point, so let me make it concrete.
Say you're a first-time manager and you've got a one-on-one this afternoon with someone whose work has quietly slipped for months. You've been putting off saying anything because you like them and you don't want to make it weird.
Ask a normal AI and you'll get a solid explainer: be specific, focus on behaviour not character, use a feedback model, create a safe space. All true. All useless to you at 2pm, because you already knew the principles. You're not stuck on the theory. You're stuck on the actual sentence to open with and the fear of how they'll react.
A skill built from Kim Scott's Radical Candor methodology does something different. It recognises the pattern — months of avoidance plus genuine care for the person is a textbook case of what Scott calls Ruinous Empathy, caring so much about the relationship that you protect someone right out of the feedback they need to grow. It names that for you. Then it writes you the actual opening line, calibrated to your situation. Then it flags the resistance you'll probably hit — they'll blame workload — and prepares your follow-up. You walk into that room with the judgment of someone who has coached this exact conversation a thousand times. That's not a summary of a book. That's the expert in your ear.
Or you're a founder and your landing page isn't converting. The generic answer is a checklist of best practices you've read before. A skill built from a real positioning methodology asks the questions the expert would ask, in their order, finds the specific thing that's wrong — you're describing what your product is instead of what it's for — and rewrites the hero section to prove the point. You didn't get told what good looks like in the abstract. You got your own page, fixed, the way the specialist would fix it.
Same model underneath in both cases. Radically different experience, because one is reciting information and the other is applying expertise.
Why the difference is bigger than it sounds
There's a reason "apply" beats "explain" by such a wide margin, and it's worth understanding because it tells you what to reach for and when.
Real expertise isn't a body of facts. It's four things working together that don't show up in any summary. It's knowing what good actually looks like — not in theory, but the calibrated standard you only get from seeing a thousand examples. It's knowing which questions to ask and in what order, so you diagnose the real problem instead of the obvious one. It's spotting the pattern a beginner misses — that this isn't a writing problem, it's a structure problem; that this isn't a feedback problem, it's an avoidance problem. And it's delivering the verdict in the specific form that's actually useful, not a vague "consider doing more."
A general AI has read about all four. It cannot reliably do any of them, because they were never really written down — they live in the expert's hands, built over years of doing the work. A skill is the attempt to capture exactly that and make it portable. When it works, you're not talking to a chatbot that's read a lot of books. You're borrowing a specialist's judgment for the length of one problem.
What actually changes for you
Here's the shift, and it's quieter than the hype but much more real.
The promise of AI was never "now you can get an explanation." You could always get an explanation. The promise — the one that's only now arriving — is that the gap between I have access to the world's expertise and I can act on it finally closes. You stop being the person who has to translate every framework to your own situation. The framework translates itself.
It means the help you used to be able to afford only at a crisis point — the $500-an-hour, two-week-wait kind — is available for the small, ordinary decisions that make up most of a working life. The one-on-one this afternoon. The page that won't convert. The bet you can't tell is good. Those never warranted a consultant. They absolutely warrant better than a generic answer.
This isn't a chatbot, a course, or a consultant. It's a fourth thing. It's the consultant's judgment with the book's reach — expertise you can download and point at whatever's in front of you. The information was always going to be free. What you actually needed was the judgment to use it. Now you can have that too.
Cheers, Adam