Start with a specific use, not the tool
Many teams try an AI assistant out of curiosity, then drop it for lack of a clear use case. Before choosing a tool, list three repetitive tasks in your team: meeting minutes, first drafts of letters, document summaries. Pick one, define what you expect from the result, and test it for two weeks. A specific use can be measured; a vague intention cannot.
Set simple rules on data
A manager’s first question is not “what can the tool do?” but “what can we entrust to it?” Information about customers, salaries, personnel files, or contracts has no place in a tool whose workings you do not control. Write a one-page set of rules, validated with your leadership or your legal department, and share it with the whole team. A short rule that everyone knows is better than a detailed policy that nobody reads.
Always proofread, and say that you proofread
AI-generated text can be fluent and still wrong: an incorrect figure, a nonexistent source, a tone unsuited to your audience. Responsibility stays with the person who signs. Establish a clear habit:
- any document produced with the help of AI is reviewed by a human before sending;
- figures and references are checked against the source;
- everyone indicates, internally, when AI was used to prepare a deliverable.
This transparency prevents misunderstandings and reassures team members who hesitate to admit they use it.
Reinvest the time saved in the human side
If AI saves you an hour on meeting minutes, decide in advance what you will do with it. One-on-one conversations, feedback, support for an employee who is struggling, or preparing for a difficult decision are activities where your presence matters. Time saved that disappears into new administrative tasks brings nothing to the team.
Listen to your team’s concerns
Some team members fear for their jobs, others feel behind on the tools. Do not dismiss either reaction. Hold a half-hour discussion where everyone can say what they use, what bothers them, and what they would like to understand. That way you will identify real training needs instead of assuming them.
A concrete example: meeting minutes
Take the most common task. Before asking an AI for a summary, specify the format you want: decisions, actions, owners, deadlines. Then review the text against your notes, correct names and dates, and send it under your own name. After two weeks, compare the time spent with before. If the gain is real and the quality is at least equivalent, extend the use to a second task. If not, drop it without regret: a trial that does not convince is part of the process.
Also keep a record of what works: the instructions that give good results, the errors encountered, the cases where the tool is not suitable. This small shared notebook saves everyone from repeating the same trials and makes the team’s progress visible.
Train to make informed decisions
Using AI responsibly requires understanding its possibilities and limits, not becoming a computer scientist. The online course Managing in the Age of AI is designed for managers who want to use artificial intelligence responsibly and profitably. To go further on a specific managerial skill, Delegating with Impact Using AI covers delegation, a core task for managers.
You can also browse our course catalog to put together a path suited to your team. Each completed course leads to a learning certificate.
Where to start this week
Choose a single use, write down your rules on data, announce them to the team, and set a review date. This first short cycle will teach you more than a long phase of upfront reflection.