AI for managers: using it yourself and leading a team that does

Managers face AI from two sides. They can use it in their own work, for planning, writing and analysis. And they lead teams whose tasks are changing because of it, which raises questions about workload, quality, fairness and trust. This page covers both, and points to our detailed guides where a topic needs more depth.

The short answer: use AI for the preparation around management, such as drafting, summarizing and analysis, but keep judgment about people with yourself. For your team, decide where AI fits in the work, set the rules on data and review, make sure people are trained, measure the results, and stay accountable for what the team and its AI tools produce.

Part 1: Using AI in your own work

Good uses for a manager:

• Preparing: agendas, briefing notes, questions for a one-to-one.
• Writing: first drafts of updates, plans and announcements.
• Summarizing: long threads, reports or feedback, checked against the original.
• Analysis: a first pass at numbers, which you verify. See checking AI output.
• Thinking it through: testing a plan, listing risks, finding counterarguments.

Keep these with you:

• Judgments about people: performance ratings, hiring, pay, discipline. In the EU, AI used for these decisions is listed as high-risk; see AI at work and workers' rights and AI in performance reviews.
• Sensitive conversations: feedback, conflict, bad news.
• Confidential information: keep it out of unapproved tools. See AI agents and personal data.

AI can produce confident but wrong output, which NIST calls confabulation (NIST AI 600-1). A manager who forwards an unchecked AI summary has made the error their own.

Part 2: Leading a team that uses AI

Decide where AI fits

Go task by task with the team: which tasks AI can help with, which it should not touch, and who checks the result. The people doing the work know where the risks are. If you are introducing agents, follow introducing AI agents at work.

Set the rules

Make sure everyone knows the approved tools, the data rules and what needs review. If your organization has no policy, push for one; see AI policy for employees.

Train people for their tasks

Different people need different training, and the people supervising agents need the most. See how to train employees on AI.

Watch workload and fairness

The OECD reports that workers often see AI improving task autonomy and enjoyment, but also report higher work intensity after AI adoption (OECD Employment Outlook 2023). Do not quietly raise targets just because a tool makes a task faster; check whether people are working harder, and whether the gains are shared fairly. If AI is used to measure your team, see AI monitoring at work. For what research finds about AI that sets the pace and its effect on health and safety, see AI and workplace safety, and if your team is unionised, AI and unions shows how some contracts handle AI.

Measure what changed

Look at quality, errors, time and how people feel about the work, before and after. For agents, use measuring AI agent performance.

Own the result

Your team is accountable for its output, including what its AI tools produce, and you are accountable for the setup. Make sure every AI system your team uses has a named owner and a clear review rule. See human in the loop and accountability.

Supporting people through the change

Some team members will be keen, some anxious, some skeptical. Talk openly about which tasks are changing, what new work it creates, and what it means for roles. Give people time to learn and credit for catching AI errors. Where roles are shifting toward supervising AI, the AI agent manager job description shows what that work involves, and digital collar skills covers what people need to grow into it.