“We need AI” is a technology in search of a problem. The right first sentence names the pain in plain words — and most companies skip it.

It usually starts with a demo. Someone sees what the software can do, the room gets excited, and by the end of the quarter there's a contract. Nobody wrote down what problem it was supposed to solve, because the problem was never the starting point — the technology was.

This is how most AI projects end up as shelfware. Not because the technology doesn't work, but because it was never pointed at anything. An AI model can't fix a process nobody described. It can only automate the confusion faster.

The right order is unglamorous. First, name the operational pain in one plain sentence: orders take too long to process, the night shift's output is anyone's guess, nobody knows where the margin goes. Then measure it — roughly is fine. Only then ask what technology fits.

Notice what happens when you do it this way: the technology choice often gets smaller, not bigger. Sometimes the answer is a dashboard, not a model. Sometimes it's a checklist. Starting with the problem keeps you honest about what you actually need.

AI is an amplifier. Point it at a clear problem and it multiplies the result. Point it at a vague wish and it multiplies the vagueness. The starting sentence decides which one you get.

So the next time someone says “we need AI,” try a different first sentence: “Here's the problem, in plain words.” Everything after that gets easier.