One expert. A codified system. An AI delivery layer underneath it. Here are the five components of the model, the measured economics behind the leverage, and the two places the whole thing breaks.
Traditional coaching and consulting share a stubborn limitation: revenue is tied directly to the mentor's personal hours. More clients means more calls, more emails, more one-on-one time — until the calendar is the ceiling on the entire business. An AI mentorship business is built specifically to remove that ceiling without removing the mentor from the parts of the relationship that actually need a human.
TL;DR — the model in one diagram, in words: at the center sits a Core (your values and non-negotiables). Wrapped around it is codified expertise — your industry judgment written down as a teachable method. That method feeds an AI clone that answers, drafts, and delivers in your voice at near-zero marginal cost. A four-hour operating system routes only the genuinely hard cases to your personal hours. And the whole structure sits inside a mentor ladder — an offer sequence and a network that move people from a cheap book to direct access. Money flows in at every ring; your hours are spent only at the center. The economics work because AI compresses the repeatable middle: the best causal evidence we have — Brynjolfsson, Li & Raymond's QJE study of 5,179 support agents — measured a 14% average productivity gain from a generative AI assistant, and 34% for novices. That is the leverage this model is built on.
Earlier versions of this dispatch described the model as four generic "layers." That was my compression, and it flattened the thing. The Mentor Economy builds the model from five named components across chapters four through eleven, and the order matters — each one depends on the one before it.
Everything starts here, and it is the component most business-model explainers skip because it doesn't look like a business asset. The book is precise about what it is:
A Core is built from three components: what you believe, who you are becoming, and what you will not compromise to win.
— The Mentor Economy, Chapter Four
Why does a values statement lead a business model? Because in a model where AI can generate infinite plausible content, the Core is the only component competitors cannot copy — and it is what clients are actually buying when they pay for a mentor instead of a chatbot. The consumer data backs this up harder than most AI enthusiasts want to admit; more on that in the limits section.
This is your two decades of pattern recognition — what actually goes wrong in your industry, in what order, and what to do about it — written down as a named, teachable method. Not a vibe, a method: stages, checkpoints, failure modes. Codification is the hinge of the whole model. Uncodified expertise can only be sold by the hour. Codified expertise can be delivered by systems.
An AI system trained on your codified method and your actual writing, so it answers client questions, drafts content, and runs follow-up in your voice, from your material. This is the component doing the economic heavy lifting, and it is the one place where we have genuinely rigorous evidence instead of vendor claims — the QJE numbers below. If you want the build details, I've written a full dispatch on building your AI clone.
The routing layer. Systems handle the repeatable 80%; a deliberate, protected block of the mentor's personal hours handles only the cases that need real judgment — the client whose situation matches no pattern, the hard call, the moment that needs someone who has actually been there. The mentor's scarce hours stop being the delivery mechanism and become the escalation tier.
The distribution and offer structure: a sequence of offers stacked by price and trust — free content, a cheap book, a course, a group, direct access — plus the network mechanic of mentors above you and mentors below you. The ladder is why the model doesn't need cold-selling a $10,000 engagement to strangers. I've broken the offer math down separately in the expertise ladder blueprint.
Here is my position, stated plainly: most "AI 10x'd my business" claims are unmeasured, but the honest, measured numbers are still strong enough to justify the model — and the strongest ones favor exactly the shape of business this model describes.
Start with the causal evidence. The Brynjolfsson, Li & Raymond study (NBER Working Paper 31161, published in the Quarterly Journal of Economics, 2025) tracked 5,179 customer support agents given a generative AI assistant. Issues resolved per hour rose 14% on average — and 34% for novice workers. Read that second number again in mentorship terms: the AI worked by disseminating the best practices of the most able workers to everyone else. The assistant was, functionally, a scaled mentor. It also improved customer sentiment and employee retention. This is the single most rigorous data point on AI in service work, and it describes expertise transfer — the exact product a mentorship business sells.
Sources: NBER Working Paper 31161, "Generative AI at Work" (Brynjolfsson, Li, Raymond, QJE 2025); Ruzuku, "The Completion Gap" (32,000+ courses, 2011–2025); Class Central MOOC median.
Now the professional-services side. The Thomson Reuters Future of Professionals Report 2025 (2,275 professionals surveyed) found professionals expect AI to save them 5 hours a week — roughly $19,000 a year in value per person — and that organizations with a visible AI strategy are twice as likely to see AI-driven revenue growth. A year later, the Thomson Reuters Institute's 2026 AI in Professional Services Report shows adoption nearly doubling: org-wide AI usage hit 40% in 2026, up from 22% in 2025. For a one-person expert business, five reclaimed hours a week is not a rounding error — it is a second product line, or the entire content layer of this model, produced inside time that used to be admin.
And the market this leverage applies to is not small or shrinking. The ICF 2025 Global Coaching Study (conducted with PwC, 10,000+ participants across 127 countries) puts global coaching revenue at $5.34 billion with a record 122,974 practitioners — up 15% since 2023. The demand side is growing; what the AI mentorship model changes is how much of it one person can serve.
Finally, the delivery evidence — the part that justifies keeping component four human. Across 32,000+ courses and 50,000+ enrollments analyzed by Ruzuku, cohort-style courses reached 64.2% completion versus 48.2% for self-paced — and courses with active community discussion hit 65.5% versus 42.6% without. The median MOOC completes at 12.6%. Programs that pair recorded content with live human coaching (altMBA at 96%, HBS Online at 85%+) sit at the top of the range. The pattern is unambiguous: content scales attention, but humans produce completion. A model that automated the human layer away would be destroying its own results.
Because AI collapses the marginal cost of the outer components, the model can afford a genuinely cheap entry offer — a book, a guide — that earns trust and filters for seriousness, while the scarce human hours are reserved for the top of the ladder and priced accordingly. The ladder isn't a funnel trick. It is an honest match between how much personal access a client needs and how much they are willing to invest to get it. Each rung up buys more of component four. The full pricing logic lives in the ladder dispatch; the point here is structural: the ladder only works because components two and three make the lower rungs nearly free to deliver.
I'd rather tell you this now than have you discover it six months in. The model has two hard failure modes, and no amount of tooling fixes either.
Break one: no real expertise. AI can draft an email; it cannot invent twenty years of judgment about what actually goes wrong in a specific industry. Businesses that skip component two and build directly on AI-generated content produce something that sounds plausible and has no depth behind it the first time a real client asks a real, specific question. The QJE study is often quoted as evidence AI replaces expertise — it shows the opposite. The AI assistant worked because it was trained on the practices of the most able human workers. Remove the experts from that system and there is nothing to disseminate.
Break two: real expertise, never codified. This is the quieter failure. Genuine experts whose method lives entirely in intuition — who "just know" — have nothing to hand the AI clone and nothing to structure the delivery system around. Every client still requires their personal hours, which means they have rebuilt the hourly consulting model with extra software costs. Codification is unglamorous writing work, and it is the actual price of admission.
There is also a market-level honesty check worth stating. Most firms claiming AI margin gains have not measured them: the 2026 Thomson Reuters Institute report finds only 18% of organizations measure ROI on their AI tools, and only a small share analyze AI's impact on revenue or client outcomes at all — so most margin claims are self-reported expectation, not measurement. And the human layer is not optional garnish: in a December 2025 SurveyMonkey survey of 2,017 US adults, 42% said they would pay extra for access to a human, 50% would cancel a service they found was solely AI-driven, and 79% prefer interacting with a human over an AI agent. The market is telling you exactly where the value sits. Build the AI underneath the mentor, never instead of the mentor.
Not with the AI. Start with component one and two: write down what you will not compromise, then write down the actual process you use, in plain language, as if explaining it to someone starting from zero. That document is the seed of the clone, the curriculum, and the ladder all at once. If you're still deciding whether this whole economy is real or a label, read what the Mentor Economy actually is first — then come back and build component two. Everything else in this model is leverage, and leverage multiplies whatever it is pointed at. Point it at something real.
No. The AI systems in this model — an assistant trained on your material, automated follow-up, content drafting — are configured through plain-language instructions and off-the-shelf tools, not custom software. The scarce ingredient is codified expertise, not engineering.
Only if the business is built badly, and the data says the risk is real: in a December 2025 SurveyMonkey study, 50% of US consumers said they would cancel a service they discovered was solely AI-driven. Done right, AI absorbs the repeatable groundwork so the mentor shows up more, not less, in the moments that matter. The AI is infrastructure; the relationship stays human.
A single low-cost offer — a book, guide, or short course — delivered with AI-assisted content and a simple automated follow-up sequence. That validates the whole model before you build anything high-ticket. The full offer sequence is covered in the expertise ladder dispatch.
Two places: when there is no real expertise underneath (AI-generated content with no lived judgment behind it collapses at the first specific client question), and when real expertise exists but was never codified — if your method lives only in your head, AI has nothing to deliver and every client still requires your personal hours.
Author of The Mentor Economy and co-founder of MentorMe. He writes about turning hard-won expertise into AI-leveraged one-person businesses.
The Mentor Economy is the full system — free, you just cover $9.95 shipping.
Get Your Free Copy