One tool can play four different parts
Many leaders judge their AI skill by the quality of a prompt. That is too narrow. A better test is whether AI helps you do four kinds of work.
1. Make a first version
Use AI to turn known facts into a draft you can react to. This could be a customer email, a short brief or three ways to explain an idea.
Give it the source facts and a good example. Do not ask it to invent missing evidence.
Useful test: Did the first version help you begin, even if you changed most of it?
2. Check your thinking
Ask AI to find a weak claim, a missing voice or a risk you may have ignored. Give it a clear role, such as a careful buyer or a finance lead.
Do not treat the answer as truth. Use it as another view to inspect.
Useful test: Did it reveal a question you now want to answer with real evidence?
3. Carry a repeatable job
Some work has the same shape each time. Meeting preparation is a good example. The sources change, but the questions and output may stay the same.
Write down the steps. Show the AI what good work looks like. Add a point where a person checks the result.
Useful test: Can another person run the job and see how the answer was made?
4. Help you compare hard choices
AI can hold several options, facts and views in one place. It can show where two choices differ and what evidence would change the call.
The leader still decides. The value is a clearer field of view.
Useful test: Can you explain the choice, the trade-off and the fact that mattered most?
Try all four on one real problem
Take a live decision. Ask AI to make a first view, question it, help with one repeated step and lay out the final options. Keep a note of where it helped and where a person had to step in.
That note is more useful than a score from a training course. It shows how you work with AI when the result matters.