Here's an uncomfortable truth for anyone who's spent a decade building a body of expertise: a chatbot can now summarise your life's work in about four seconds, for free, and nobody asked your permission.
Your book. Your framework. The thing you refined over fifteen years of doing the actual work — distilled into a tidy paragraph the moment someone types your name into a prompt box. No attribution. No royalty. No acknowledgment that the only reason the model can say anything useful about your field is that you, and people like you, wrote it all down first.
This is not a hypothetical. In April 2025, researchers published evidence suggesting OpenAI had trained its models on paywalled O'Reilly books — without a licensing agreement in place. O'Reilly's library is 45,000-plus titles from more than 200 publishers. That's not a rogue PDF someone uploaded. That's a substantial chunk of the world's professional and technical knowledge, apparently absorbed into a commercial product whose owners never cut the authors a cheque.
I don't raise this to start a fight about copyright law. That fight is happening, it'll grind through the courts for years, and the lawyers will be fine. I raise it because it exposes something every expert needs to understand about the position they're now in. Your expertise has already been consumed. The question is no longer whether AI knows about your work. It does. The question is whether you ever see a cent from it.
"Knowing about" is not "being able to do"
Here's the part that should give you hope, because it's where the whole thing turns.
There's a world of difference between a model knowing about your methodology and a model being able to apply it. Ask a base AI to explain your framework and it'll give you a competent summary — the bullet points, the history, the gist. Ask it to actually run your framework on a real, messy, specific problem and it falls back to generic advice. It knows the words. It doesn't have the judgment.
I've tested this more times than I can count. We took Ryan Singer's Shape Up methodology and put base Claude up against a properly extracted version of the same thinking. On a real product scenario — a migration that had been stuck for two weeks — base Claude proposed a new architecture and predicted it would fail. The extracted version spotted the actual problem: the requirements themselves were contradictory, and no amount of redesign would fix that. One knew about Shape Up. The other could do Shape Up. Across our scoring it was an 18% lift overall, but the gap was concentrated exactly where you'd expect — voice, mastery, methodology. The expert dimensions.
So the training-data situation is genuinely double-edged. Yes, the labs have hoovered up the explanation of your work. But the explanation was always the cheap part. The valuable part — the calibrated judgment, the "I know what good looks like because I've seen a thousand of these," the diagnostic instinct that tells you the problem is upstream — that didn't make it into the training data. It mostly lives in your head, and in the tacit space between the lines of whatever you published. That's the part nobody's paying you for yet. It's also the part nobody can take without you.
The value chain has a hole in it
Step back and look at how the money moves in this new arrangement, because it's genuinely odd.
You create the knowledge. The publisher or the platform distributes it. The AI lab ingests it. The end user gets the benefit — a synthesised, instant, free answer drawn from your thinking and ten thousand other people's. Every link in that chain captures value except the one at the very start. The labs raise billions. The chatbots get more useful by the month. And the experts whose actual expertise makes the whole thing work are sitting outside the building wondering why their book sales are soft.
This is the bit that bothers me, and it should bother you. Not because AI is evil — I build with it every day and I think it's the most important tool of our lifetimes — but because the economics are simply incomplete. There's no mechanism for the expert to participate. The system was built to consume published knowledge, not to compensate the people who produced it. That's not a moral failing of any one company. It's a missing piece of infrastructure.
And missing infrastructure is an opportunity, not a tragedy.
The alternative: package it, protect it, get paid for it
What if your methodology didn't just sit in a book waiting to be scraped? What if it was a product — one that applies your actual judgment to someone's real problem, that you control, and that pays you every time it's used?
That's the thing I've spent the last year building, so I'll be straight that I'm not a neutral observer here. But the shape of the answer matters regardless of who builds it.
You take your methodology and you extract it properly — not a summary, the real decision logic. The thresholds you use to judge whether something's good enough. The questions you ask, in the order you ask them. The patterns you spot that a beginner walks straight past. The specific way you present a verdict. That gets packaged so an AI can apply it faithfully, in your voice, to whatever the user brings.
Then — and this is the part that actually protects you — it gets delivered in a way that never hands your methodology over. There are two ways to ship it. One is open: a readable file anyone can inspect, good for building trust and reach. The other is premium, and this is the one that matters for valuable IP: the methodology lives on a server and is never sent to the user or the model wholesale. The user gets the output of your judgment. They never get the judgment itself to copy. Your IP is protected by architecture, not just by a terms-of-service page nobody reads. The expertise leaves the building as advice, never as the source.
And you get paid for it. Properly. The model I landed on is a 70% revenue share to the creator, because the methodology is the entire product — everything else is plumbing. Every time your skill is used, you earn. Not a one-off advance against royalties you'll never see again. A line that pays out while you sleep, scaling with use rather than with how many hours you can personally bill.
This is a market, not a side hustle
If your instinct is that this sounds small — a neat way to make a bit of passive income off an old book — look at the numbers on the creator economy and recalibrate.
The creator economy was worth around $56 billion in 2024 and is projected to reach over $320 billion by 2032. That entire market was built on people packaging what they know and selling access to it — courses, newsletters, memberships, communities. What's changed is the delivery mechanism. For twenty years, "package your expertise" meant record a course and hope people watch it. The expertise sat there, static, waiting for the user to do the work of applying it to their own situation. Most never did. That's why course completion rates are a punchline.
An AI skill closes that gap. It doesn't ask the user to translate your framework to their problem. It does the translation. That's a fundamentally better product than a course, sold into a market that has already proven it will pay for expertise. The experts who work this out early — who get their methodology packaged, protected, and earning before everyone else catches on — are going to look, in a few years, like the people who bought the good domain names in 1998.
What I'd do if it were my life's work
So here's where I land, and I'll say it plainly because hedging helps no one.
Your expertise is already in the machine. That ship has sailed, and no amount of being cross about it will sail it back. But the machine only got the cheap half — the explanation. The expensive half, the judgment, is still yours, and for the first time there's a way to package that judgment into something people and their AI agents will pay to use, without ever giving it away.
The experts who treat this as a threat will spend the next few years watching their work get summarised for free and feeling vaguely robbed. The ones who treat it as a distribution channel — the best one they've ever had — will build an asset that earns for them across every situation their expertise touches, at a scale no consulting practice or speaking calendar could ever reach.
I know which group I'd want to be in.
Cheers, Adam