Don't Start With AI

Every week I talk to entrepreneurs who want to build something with AI. The first question they almost always ask is: what can I build with AI?

I understand the instinct. The capabilities are extraordinary and expanding quickly, and it is natural to look at what the technology can do and try to find a use for it.

But I think it is the wrong starting point.

My Advantage Was Not Technical

I am not an AI engineer. I have never built a large language model, and I do not come from the technology industry. My background is in marketing and small business — years of working directly with owners, understanding their problems, watching the same challenges repeat across thousands of different businesses.

My advantage, such as it is, did not come from knowing more about AI than the people who build it. It came from knowing a problem.

I had spent years watching the same gap appear in conversation after conversation: good marketing requires more expertise, coordination, and sustained effort than most small businesses can afford. The work exists. The need is real. The pain is genuine. And the traditional solutions — agencies, in-house teams, marketing directors — were priced for companies ten times the size of the businesses I was working with.

I understood that problem with a depth that only comes from years of direct exposure to it. So when AI began acquiring the capabilities to solve it — when agentic systems made continuous, coordinated, intelligent marketing work seem genuinely possible — I could see it clearly.

Not because I was ahead of the AI curve, but because I was ahead of the problem curve.

Start With the Problem Curve

That is the advice I would give any entrepreneur thinking about building in AI: do not start with the technology. Start with a problem you understand extremely well.

AI is advancing so quickly that remarkable new capabilities appear almost constantly. Somewhere inside those capabilities may be the solution to a problem you have understood for twenty years.

The question worth asking is not what can I build with AI, but what problem do I understand well enough to recognize when AI finally becomes capable of solving it.

That is where the real opportunities are hiding.