A better first question for AI.

“Where could AI help us?” is a useful opening. A more specific question can make the next conversation much easier: “Which part of our work deserves a closer look?”

That shifts the discussion toward something the team knows. A recurring task. A frustrating handover. A document that is difficult to start. A decision that needs clearer information.

Make the task visible.

Choose one example and describe how it works today. Who starts it? What do they need? What does a useful result look like? Where does someone spend time checking or correcting it?

You may discover that the first improvement is a better brief or a clearer process. You may also find a specific step worth exploring with AI. Both are useful outcomes.

Give the experiment a boundary.

Work with a suitable, non-sensitive example. Define the task narrowly enough that a person can review the result. Agree on what you are looking for before you begin: clearer wording, a useful structure, or a reasonable first set of options.

Keep the original approach available for comparison. A result that looks polished still needs to be checked against the purpose of the task.

Finish with a decision.

After the experiment, ask the team what improved, what became harder, and what still required judgment. Decide whether to try again, adjust the approach, or leave the task as it is.

The aim of a first experiment is to learn something specific enough to guide the next one.