What AI Can’t Commoditize: The Four Things That Still Defend a Business

Two things became more valuable over the last two centuries for the same strange reason. When the camera arrived, painting did not die — the best of it got more valuable. When the motor engine arrived, the horse did not just lose its job — riding became something people now pay a premium to do.

There is a real principle underneath that, and it is the most useful lens you can carry into the next decade: when an input becomes abundant, value migrates to its scarce complement. Cheap photography made the eye scarce. Cheap horsepower made the experience scarce. Cheap intelligence — which is what AI actually is — makes something else scarce. The entire question is: what?

First, kill the comforting answer

Everyone wants to believe that as machines take over, humanness becomes the differentiator. It is a comforting thought. It is also where most people stop thinking — and the stopping point is a trap.

Here is the disanalogy that breaks it. A motor engine never pretended to be a horse. It replaced what the horse did, but it never impersonated the horse. AI does the opposite. It is getting good at simulating the exact thing people call the moat — empathy, conversation, personalization, the feeling of being understood. The cheap substitute for humanness is humanness-shaped. So at the low end, “the human touch” is itself being commoditized.

Narrative is going the same way. Anyone can now generate infinite copy, infinite hooks, infinite identity angles. Narrative production is getting cheaper and noisier by the month. And at the same time, verification is democratizing — reviews, receipts, screenshots, AI fact-checkers — so the half-life of a hollow narrative is collapsing. A story with nothing underneath now gets exposed faster than ever.

The one genuinely scarce act

Strip away the comforting answers and one thing is left standing — and it is the most AI-resistant skill there is: finding and framing the right problem.

AI is superb at solving a well-specified problem and nearly useless at deciding which problem is worth solving, or reframing a problem so it becomes solvable at all. Problem definition requires context, stakes, taste, and a point of view about the world — exactly the things a model does not have. Hand a model a sharp question and it is brilliant. The scarce act is producing the sharp question.

The scarce skill is no longer answering the question. It is knowing which question is worth asking.

The four-leg moat

So it is not IP or humanness or psychology. Those are arguments people have because each one is partly right. The truth is that durable defensibility is a loop of four things — and AI eats any leg you remove.

Remove substance and you get ignored. Remove narrative and you stay invisible. Remove proprietary insight and you get copied. Remove the trusted anchor and you get disintermediated. Each leg covers a different failure mode.

The discipline rule

If you take one operating rule from this, take this one: substance earns the right to narrative. Never run narrative as a substitute for substance. In a world where everything can be verified, that ordering is the whole game — and it is the line that separates a real brand from an AI-generated one with nothing underneath.

From conglomerates to kiranas — but not uniformly

It is tempting to declare this a universal law: humanness wins everywhere, from the largest conglomerate to the corner kirana. It does not. The kirana is the most human-touch retail format in India — the owner knows your family, remembers your order, extends informal credit — and it is still losing share to ten-minute quick-commerce, which has zero humanness. At the commodity tier, convenience beat relationship.

The lesson is that this is category-elastic, not universal. Narrative and identity leverage is enormous in beauty, apparel, food, and the creator economy — categories bought with emotion and identity. It is weak in infrastructure, logistics, and compliance — categories bought on reliability and spec. Know which game you are actually in before you decide which leg to lead with.

So the move is not “be more human” or “tell a better story.” It is sharper than that: find a problem worth solving that others have mis-framed, solve it with real substance, make the value legible through narrative grounded in truth, and anchor all of it to a name people trust. Do all four, in that order, in a category where they matter.