How to Tell If an AI Tool Solves a Real Problem or Just Sells You One

There is a lot of money being made right now selling small businesses solutions to problems they did not have.

I build AI tools for a living, so I am not against them. I am against you paying for one that changes nothing. These are the questions I would ask.

One: what did I do about this yesterday?

If the answer is nothing, because it was not a problem, stop there.

The best tools fix something you are already doing badly, slowly or late. If you have to invent a use for it after buying it, you bought a solution looking for a problem.

Two: what happens on a Tuesday?

I use this constantly and it kills most purchases in one sentence.

Describe your Tuesday afternoon with this thing working. If the description is "we would be using AI", that is not an answer. If it is "I would not spend two hours writing quotes", that is.

Three: does it create work for somebody?

A lot of these tools shift effort rather than removing it. The output now needs checking, the exceptions still come to you, and someone has to keep it fed.

That is not automatically bad. But if a person now spends forty minutes a day reviewing what it produced, that is forty minutes, and it belongs in the price.

Four: how wrong is it allowed to be?

Ask what happens when it gets something wrong, because it will.

If it is drafting internal notes, occasional nonsense is survivable. If it is quoting prices to customers or touching anything financial, it is not, and the honest answer to how much it costs to make something reliably right is usually several times the demo price.

Anyone who tells you it does not make mistakes is either lying or has not used it much.

Five: can I leave?

Where does your customer data live? Can you export it? What happens to the thing you built if they double the price or shut down?

This is the question people skip and regret. A tool is something you use. A dependency is something that uses you.

The pattern worth knowing

The tools that work for small businesses are almost always narrow and boring. They do one job that happened eleven times today and will happen eleven times tomorrow.

The ones that disappoint are broad and impressive. Platforms. Suites. Anything demonstrated with a dashboard rather than a task.

That is because your business is specific and a platform is general. Whatever makes the demo look good is usually the thing that makes it a poor fit for you, which is more or less the case for building around what your business actually does.

What I tell people who ring me

Half of them do not need me. Some need one subscription and an afternoon. Some need their offer fixed before any tool will help.

I would rather say that than take the money, and I have said it enough times to have lost work over it. The ones who stayed have stayed for years, which tells you which approach actually pays.

If you are weighing up whether to bring anyone in at all, is an AI consultant worth it for a small business is the same honesty applied to hiring me.

If you have been pitched something and want a second opinion before you sign, get in touch. I will tell you honestly whether it is worth it.