Can AI grow a business that isn't clear on who it is?

Mostly, no. AI multiplies whatever architecture it's pointed at. A business with clear positioning gets sharper and faster. A business that hasn't decided who it serves gets more generic output, produced quicker, at a lower cost per unit of sameness. AI can run the engine. It can't decide what the engine is for.

Are you actually behind on AI?

Picture the move almost everyone has made by now. You open a chatbot, type something like "write my marketing strategy," and hit enter with a small flicker of hope. The answer comes back fast. It reads fine. Competent, even.

Then the hope leaks out. Because the same answer is landing on the screen of the agency three towns over, the coach down the street, and every other owner who typed the same tired prompt that morning. Fine, and completely interchangeable.

It's an easy move to make, and not a lazy one. You've been told, on repeat, that you're behind on AI, that the people who win in 2026 are the ones with the right prompts and the right system, and that if you bolt the tool on fast enough the leads will follow. That message is everywhere right now. The loudest voices in the coaching world have turned "you're behind on AI" into a product, the way they once turned "you need a funnel" and "you need a proven system" into products.

You're probably not behind on AI at all. You're behind on a decision that's easy to keep putting off: who is this business actually for, and what does it do for them that the next guy doesn't. No tool decides that for you. It can't. It will happily generate around the hole, though, and at remarkable speed.

What is AI actually multiplying?

Think of AI as a multiplier, not a source. You bring the input. It scales the output.

If the input is a clear point of view, AI takes that point of view and spreads it cleanly across a year of emails, posts, proposals, and pages, in your structure, at a pace you could never hit by hand. If the input is a shrug, it scales the shrug. You get a year of competent, forgettable content that could carry anyone's logo.

This is the architecture problem wearing a 2026 costume. We say it to founders constantly: most of the time you don't have a marketing problem, you have an architecture problem. The business was built around the founder, on relationships and instinct, and the underlying structure of who-we-serve and why-us was never actually written down. It lived in your head. You could improvise it on a sales call. AI can't read your head. It reads your prompt. And a vague prompt is just your unwritten architecture, exposed.

The fix is smaller than it sounds, and it has nothing to do with finding a better tool. Write down the one kind of client you're better for than anyone else, and why. Feed that to the same chatbot that was handing you oatmeal an hour ago. The tool never changes. What you point it at does, and the output changes with it.

So what about the people selling AI as your operating system?

There's a whole genre of advice right now that sells AI as the new operating system for your business. Install it, learn the prompts, and the whole thing finally runs itself. I'll be fair to it, because some of the tactical stuff is actually useful. But I've seen this movie before. A few years ago the funnel was going to be the operating system. Before that it was the guru's framework, the folder system, the playbook you bolted on so you could stop carrying the business in your head. Same promise every time: here is the system, install it, quit having to think so hard.

AI is the most convincing version yet, because on the surface it produces. You type, it answers, and the answer looks like progress.

But an operating system only runs the programs you feed it. The programs still have to exist. Your business already runs on one, and no tool can be it for you. That operating system is your architecture: who you serve, why they choose you over the obvious alternative, and how positioning, marketing, sales, and retention actually connect. AI runs on top of that layer. Feed it a clear one and it hums. Feed it a missing one and you've automated the guesswork, faster and more fluently than you used to.

The crowd selling AI as the answer stays quiet about that layer. "Decide who you actually are" has never fit on a sales page as neatly as "use my prompts and win."

Why do some businesses get sharper with AI and others blander?

This isn't just my hunch from client work. A Harvard Business School and Boston Consulting Group field experiment in 2023 put 758 consultants to the test with and without AI. On tasks that sat inside what the AI was good at, performance jumped more than 40 percent, and the weakest performers gained the most. AI lifted the floor.

But the researchers also handed the consultants a task that looked similar and quietly sat outside the tool's real competence: a business problem that needed judgment about a specific company. On that one, consultants using AI were about 19 percentage points less likely to reach the right answer than the ones working without it. The tool produced something confident, fluent, and wrong, and smart people followed it off the cliff.

Read those two results together and you have the whole story. AI raises the floor for everyone, which means the floor stops being an advantage. And on the work that actually requires deciding who you are and what you'd stake your reputation on, a confident tool pointed at a fuzzy business doesn't rescue the thinking. It launders the confusion into something that reads well.

That's the part the "you're behind on AI" crowd skips. They're selling you a faster floor. They're quiet about the fact that everyone else just bought the same one.

Why does sounding like everyone else cost more now?

For most of the last decade, the scarce thing was production. Making the email, the post, the proposal took time, so doing more of it felt like progress. That scarcity is gone. Anyone can produce endlessly now, for almost nothing.

When production goes to zero, the price of being generic goes up, not down. If your buyer can get a competent answer from a machine in four seconds, a competent answer from you is worth roughly four seconds of their attention. The only thing that earns more is a point of view they can't get anywhere else. Rory Sutherland makes a version of this argument in Alchemy: the opposite of a good idea can also be a good idea, and the things that actually work are usually the distinctive ones, not the reasonable-sounding average everyone converges on. Average is exactly what a prediction machine produces by design. It's trained to give you the most likely next word. The most likely anything is, by definition, what everyone else also gets.

So AI didn't make clarity optional. It made the lack of it more expensive. Quietly, every week, in attention you don't earn and replies you don't get.

I'm not anti-AI. I use it daily, and on a clear brief it's the best writing partner I've had. The conviction underneath our whole practice still holds, though: there is no proven system you can bolt on to skip the thinking. There never was. AI is the most convincing version of that old lie yet, because it ships the skipped thinking as a polished document. Yikes.

Try this before your next prompt

Pick your last five pieces of AI-assisted content. An email, a few posts, a proposal, whatever you've shipped.

Read each one and ask a single question: could a direct competitor publish this word for word and have it still be true for them? If the answer is yes for most of them, you haven't found an AI problem. You've found the gap in the architecture, and the tool has been faithfully scaling it.

Then do the slower, harder thing. In one sentence a twelve-year-old could repeat back, write down who this business is for and what changes for them because they chose you over the obvious alternative. Not a tagline. A decision. If you can't write it yet, that's the work this week, and no prompt will do it for you.

Once you have that sentence, paste it at the top of every prompt you give the machine for the next month. Watch how fast "generic" turns into "that sounds like us." Same tool. Different thing pointed at it.

Sources

Frequently asked

Will AI replace my marketing strategy?

No. AI runs faster on whatever strategy already exists, and produces generic work where one is missing. It's an execution multiplier, not a decision-maker. The choice of who you serve and why they should pick you is judgment about your specific business, and that's the positioning work AI can't do for you.

Why does my AI-generated content sound generic?

Because the model is built to predict the most likely next words, and the most likely anything is what everyone else also gets. Generic output is usually a sign your inputs are generic. When the prompt carries a clear point of view about who you serve, the output sharpens. The fix is upstream of the tool.

Should I wait until my positioning is clear to use AI?

No, use it now for the work that's already clear. Just don't expect it to manufacture the clarity you haven't built. Pointing AI at a defined part of the business compounds it. Pointing it at the undefined parts produces confident filler. Sort the two, and use the tool where it actually helps.

Is an architecture problem the same as a positioning problem?

Positioning is one room in the house. An architecture problem is the foundation underneath all the rooms: positioning, marketing, sales, and lifetime value, and whether they're built to work without the founder in every seat. Weak positioning is often the first crack you notice, but the structural question is bigger than any single domain.

Ready to look at the architecture honestly?

If your AI output keeps coming back generic, the tool probably isn't the problem. The thing it's pointed at is. That's a fixable problem, and naming it precisely is where we start.

Book the conversation. We'll tell you what we see, and whether the work we do fits where you are.

Book the conversation →