AI is better than humans at a lot of strategy work. It can review an entire competitive set in minutes, compare every website, campaign and claim, hold years of brand expression in context, and work through draft after draft without fatigue. Anyone arguing otherwise is arguing with reality. Coverage is becoming abundant, and machines are objectively better at it.
The mistake is assuming that strategy is coverage. It isn't. Most companies already have more information than they know what to do with. Strategy begins when someone has to decide what matters, what doesn't, and which opportunity is worth excluding all the others for. Analysis informs that choice. It doesn't make it.
That's where AI struggles. Large language models are designed to produce the most plausible answer. Good strategy is rarely the most plausible answer. A position is a bet. It creates focus by rejecting alternatives. Ask a general-purpose model for positioning and it will usually return a polished version of category consensus. The average of everything it has seen. The problem is that consensus is exactly where no brand stands out.
The second issue is that AI has no stake in the outcome. It will confidently support almost any direction the prompt leans towards. It will often agree with the founder more than it should. It will build a compelling case for ideas that sound attractive but don't survive contact with the market. Real strategy sometimes means telling people things they don't want to hear. The market doesn't care about your vision. Customers don't value the differentiator you thought they would. Your competitors already own the territory you're trying to claim. Those conversations matter because they're uncomfortable. A system optimised for helpfulness is not naturally built to have them.
There's also the problem of internal reality. Many strategic failures aren't caused by bad positioning. They're caused by a gap between what a company says and what it can sustain. That evidence doesn't live in market reports or competitor websites. It lives in leadership teams, operating constraints, incentives, culture and execution. No model can reliably infer that from public information alone.
So the question was never whether AI can replace a strategist. The better question is what sits between the AI model and the recommendation. What evidence was gathered? What method forced a choice? What challenged the obvious answer? Without those things, AI produces category-average thinking at remarkable speed. With them, it becomes an extraordinary strategic instrument. The technology isn't the differentiator. The methodology applied to it is.