How to work better with an AI?
Published
Give a role, a clear goal, ask for a plan and reasons — then always review. You don’t need to be a prompt expert.

ChatGPT, Gemini, Claude, Copilot, Muse, Grok… whatever the tool, the quality of the answer depends a lot on how you work with it.
You don’t need to become an expert in prompt engineering.
Here are four simple tips.
1. Give it a role
Tell the AI what role it should play to help you.
“You are a strength and conditioning coach specializing in 15- and 16-year-old athletes.”
That immediately gives it context and helps it adapt its answers.
2. Give it a clear goal
Explain what you want to achieve.
“I want to improve these athletes’ endurance and strength over a 12-week period.”
The clearer the goal, the better the answer can fit.
3. Start by asking for a plan
For a complex task, don’t ask for everything at once.
“Prepare a plan to help me design the workouts.”
Look at the plan, adjust it if needed, then move forward step by step.
4. Ask it to justify its answers
Simply add:
“Explain and justify your recommendations.”
You’ll understand its choices better — and you’ll be able to question them.
And above all: review!
An AI can give a very convincing answer… and still be wrong.
It can invent a fact, misunderstand the request, or make a bad recommendation. People often call that a hallucination.
Whatever tool you use, keep your judgment and verify important information.
The recipe
You can simply write:
“You are a strength and conditioning coach specializing in 15- and 16-year-old athletes. I want to improve their endurance and strength over a 12-week period. Start by proposing a work plan. Explain and justify your recommendations.”
A good way to start working with an AI:
A role + a goal + a plan + explanations.
And always: reread, question, and verify.