AI isn’t replacing consultants. It’s replacing course creators. Those are very different businesses. Most people haven’t noticed the difference yet.
Services beat courses in the AI era because AI delivers the information. It cannot apply it. The value of a practitioner has never been what they know. It has always been what they do with what they know, in a specific context, for a specific client, in real time.
AI closes the gap on the first part. It does not touch the second. That’s the whole argument. Everything else is context.
What did AI actually change about information delivery?
Everyone told you to build passive income. Create once, sell forever. The information products industry built an entire identity around that promise. What nobody said is that AI just destroyed the “create once” moat.
Here’s what nobody tells you: AI did not invent zero-cost information. It extended it. Once something exists digitally, copying it costs nothing. That has been true for decades. The course model was built on this structure. Real innovation at the time. Information was hard to reach, harder to produce. Real expertise meant real production time. That gap was the moat.
This is the core economics of information goods: very high fixed cost to produce the first unit, effectively zero cost for every unit after. A blog post, a PDF, a recorded video course. Produce once, sell forever. The model made sense when the production itself required years of accumulated expertise to get right.
What AI changed is not the principle. What it changed is the production side. Creating that first unit, the frameworks, the guides, the step-by-step systems, used to require years of expertise to build credibly. That expertise was the moat. The course model had one thing going for it: production was hard. AI absorbed it. A well-prompted AI can produce credible information content in a fraction of the time, across almost any domain where practitioners have historically sold courses.
The scarcity that gave courses their value was never the distribution. It was the production of the content. AI eliminates that scarcity. You cannot compete on information delivery when the best available information is free to anyone who asks.
Why the course model is losing ground
Courses sell a promise: learn this, do this, get this result. The delivery mechanism was information transfer. Buy the course, absorb the content, apply it to your situation, get the outcome.
There was always a missing step. And the industry knew it. Completion rates for online courses have historically stayed below 10%. Not because the information was wrong. Because applying a general framework to a specific situation requires something a recorded module cannot provide: judgment in context. The gap between “I understand the framework” and “I actually implemented it correctly for my situation” is the entire service business. That gap never closed.
The 5-step proprietary system. The exclusive acronym. The bonus live Q&A that happened twice in 2023. AI absorbed all of it. Turns out there wasn’t that much moat there. The system was information. AI is better at delivering information. And AI is available at 11pm on a Sunday, never has an off day, and costs $20 a month.
If you’re selling a course right now, you’re competing with a tool that works at 11pm, never has off days, and costs $20 a month. That is not a pricing competition you win. Not this decade. Probably not ever.
The completion rate problem was never a bug. It was telling you something. Information alone does not produce transformation. It never did. We all knew that. We just kept selling information. Courses built a business on the information layer while the transformation layer, the actual mechanism of change, stayed with the practitioner. AI has taken the information layer entirely. What remains is the work that always actually mattered.
Can AI replace an experienced consultant or coach?
Most arguments about AI either overreach or understate here. Let me be honest with you: AI will not replace experienced practitioners. Not because AI is limited in some general sense. Experienced practitioners are not primarily delivering information. They are applying judgment to context. Those are different products.
Research by Brynjolfsson, Li, and Raymond published through the National Bureau of Economic Research examined how AI-based assistants affected workers across skill levels. The results were specific: AI produced a 34% productivity gain for novice and low-skilled workers. For experienced and highly skilled workers, the impact was minimal.
This is what practitioners already know intuitively. The NBER research just confirmed it with numbers.
AI raises the floor. It does not raise the ceiling. Those are different things. AI functions as a knowledge-sharing mechanism. It distributes best practices from top performers to lower performers. Novice output gets better. The gap between a novice and an expert narrows. But the ceiling doesn’t move.
Think about it. AI can bring a novice’s output closer to a trained baseline. It cannot replicate what an experienced practitioner does with a specific client’s specific situation: the accumulated pattern recognition, the calibrated judgment about which path to take when several look viable, the read on what the client actually needs versus what they say they want. That is not information. It is intelligence applied in relationship. AI does not have a client. You do.
What applied expertise offers that AI cannot replicate
You’ve spent years building something AI cannot replicate. That’s not a consolation prize. That’s a structural advantage. The question is whether you understand it clearly enough to price and position it correctly.
What does a client actually pay for when they hire a consultant or coach? Not information. The internet has always had that. They pay for application. The practitioner takes their specific situation, its constraints, its history, its relationships, its competing priorities, and produces a result tailored to that exact context.
A course teaches someone what questions to ask. A practitioner knows which answers matter for this client, right now, given what has happened in the last three sessions. That accumulated context, held by a person and applied with judgment, is not information. It is intelligence. The distinction matters because it determines which business model survives an economy where the information layer has been commoditised.
If you have a real methodology with real results, you are not the one who got replaced. You are the one getting stronger. The more specific and complex the situation, the less AI can substitute for it. That is a defensible position. Selling information when AI gives it away is a pricing competition no human wins. Building a practice around applied intelligence in context is a market the practitioner owns by definition.
This is what Intelligence as a Service means in practice: the methodology is the product, the AI is the delivery mechanism, and the practitioner’s judgment is the layer no system replaces.
How do the economics of services change in an AI era?
The course model at its best: one-to-many information delivery at high margin. The structural weakness: completion rates below 10%, commoditisation risk, and constant pressure to create new content to justify the next renewal. Its defensibility came from the practitioner’s time to build it. AI eliminates that defensibility.
The service model had the opposite problem. High value per client. Hard ceiling on how many clients you could actually serve. Your presence was required for every unit of delivery. You could not scale without hiring a team, which just moved the bottleneck. AI changes that constraint.
“Scaling services and client-based businesses used to be hard or nearly impossible without a big team and lots of complexity. For the first time ever, that’s not the case. AI has changed that. We now have Intelligence as a Service.”
Here’s the shift. Courses: information delivery faces zero-cost competition. That is permanent. Services: the practitioner had to be present for every unit of delivery. That is now solvable. Load your methodology into a platform once. Apply it to each client in their own isolated workspace. Deliver at volume without rebuilding context from scratch for every engagement.
The practitioners who see this clearly are not choosing between scale and quality. They are building a structure where both exist at once. That was not possible before AI-native service delivery existed. It is now. What that looks like in implementation is a separate question. The economics are the starting point.
How to build a service practice AI makes stronger
Three shifts. Each one is a frame change, not a tactic.
From teaching to applying. The deliverable is not the framework. The deliverable is the framework applied to the client’s situation. That is a different product entirely. It cannot be pre-recorded, replaced by a chatbot, or undercut by a competitor with better production values. It requires your methodology meeting a specific client’s specific context. Anything you can teach in a module, AI can deliver in a prompt. What AI cannot deliver is the application in context. That is the product. Stop selling the map. Start selling the navigation.
From broad to specific. Courses survive at scale by appealing to the widest possible audience: broad topics, general frameworks, results that technically apply to everyone and therefore fit no one precisely. Services survive by serving a narrowly defined client at a level of specificity no course can match. Specialisation is the structural defence against AI substitution. The more specific and contextual the expertise, the less any general tool can substitute for it. Niching down used to feel like leaving money on the table. In an AI economy, it is the only position that holds.
From information to outcomes. Clients who buy courses are buying information and hoping it leads to outcomes. Clients who hire practitioners are buying outcomes directly. Pricing based on outcomes, not hours and not modules, is the most defensible position in an AI economy. It makes the result the product, not the content that gestures toward it. It also changes the conversation from “how much does this cost” to “what is the result worth.” That is a better conversation to be in.
Who this applies to and who should not pivot yet
Most people want to skip this section. That is exactly why it is here.
This argument works for practitioners with a documented, proven methodology. If you are a consultant, coach, or agency owner with a framework that has produced consistent results across multiple clients, you are already in the position the market is moving toward. AI makes your defensibility clearer and your delivery more scalable. The shift is in your favour.
You should not pivot to services if:
You do not yet have a proven methodology. Services built on untested frameworks produce worse outcomes than courses, not better. Courses can be padded with volume and production value. A consulting engagement either delivers results or it does not. If you cannot point to consistent outcomes across several clients, the methodology is not ready. Do not pivot until it is.
Your expertise is primarily in domains where AI already performs well. General productivity frameworks, broad business advice, generic marketing playbooks: these are areas where AI produces material of comparable quality at zero cost. The services advantage holds where application requires accumulated judgment that general tools cannot replicate. If the work is primarily procedural and information-dense, the service model faces the same commodity pressure courses do.
You are switching to escape a struggling course business. Services require client acquisition, delivery infrastructure, and consistent results. Pivoting because course sales have slowed does not resolve the underlying problem if that problem is the offer, the positioning, or the methodology itself. Services amplify a strong offer. They do not rescue a weak one.
You primarily want scale without client relationships. Services are relationship-intensive by design. A practitioner who wants arms-length, asynchronous delivery to a large audience may find that the course model, even under pressure, is still the better fit. The AI era does not make every business model worse. It makes the information-only model harder to defend. That is not the same thing.
The practitioners who benefit from this shift are those who are already good at what they do. AI gives them a structural advantage they did not have before. It does not create a path for practitioners who are not yet good enough to compete. The market rewards expertise. You have to have it first. Everything else is noise.
Client Intelligence is the applied intelligence platform built for practitioners who are ready to structure their methodology as a scalable delivery system.
For more on how service practices are being built in an AI economy, see the Client Intelligence blog.