8 AI Prompts for Managers: Feedback, Reviews and 1:1s

The best AI prompts for managers give the tool real evidence (what happened, when, and the impact) and ask for a draft you will rewrite, not a finished message. Use them to turn rough notes into clear feedback, structure fair performance reviews and plan sharper 1:1s, then edit every line so it sounds like you.

Below are eight copy-ready prompts, including one for tough feedback and one for your own self-review. Each works in ChatGPT, Claude or Gemini. For prompts that cover the rest of your workday, see our 50 copy-paste prompts for work.

A quick note on terms: a 1:1 is a regular private meeting between a manager and one team member. Americans usually call it a one-on-one, while in the UK you will often hear one-to-one.

How should managers use AI for feedback and reviews?

Most advice about ChatGPT for managers focuses on speed. The bigger win is clarity. AI is good at structure, neutral wording and spotting vague phrases. It is bad at knowing your people, your history with them and what they need to hear. So follow one rule: AI drafts, you edit.

Four habits make that rule work:

1. Feed it evidence, not opinions. Dates, examples and outcomes produce specific drafts. "She's great" produces mush.
2. Protect people's privacy. Use initials or roles instead of names, and never paste health details, pay or anything from a disciplinary file. Check your company's AI policy first.
3. Ask for a draft, then rewrite it. Read the result aloud and cut any sentence you would never say in person.
4. Keep the decisions yours. Ratings, pay and anything that could become a formal process belong to you and HR, not a chatbot.

On privacy, know your settings. On personal ChatGPT plans you can switch off "Improve the model for everyone" under Settings, then Data controls, according to OpenAI's guide to ChatGPT data controls. OpenAI also says it does not train on content from Business or Enterprise workspaces by default, so an approved company account is the safer place for people topics.

If you are still choosing a tool, compare the leading AI assistants first. Here is a quick map of which prompt to reach for:

1. Turn rough notes into specific feedback

Vague feedback ("great job," "be more proactive") is easy to give and hard to act on. This prompt uses the Situation-Behavior-Impact (SBI) model, developed by the Center for Creative Leadership, to turn your scribbled notes into something specific.

You're helping me, a team manager, write feedback for a direct report. My rough notes: "Tuesday client call. M. stayed calm with an angry customer, offered the refund option before I stepped in, and the customer renewed." Rewrite this as spoken feedback using the Situation-Behavior-Impact format. Keep it under 80 words, warm and plain, with no corporate jargon. Then suggest one question I can ask M. to make the conversation two-way.

Generic AI feedback examples usually come from generic prompts. Specific notes in, specific feedback out. Swap in your own moment and keep the three-part structure.

Who it suits: new managers, and anyone who wants to give more praise without it sounding hollow.

2. Prepare tough feedback without losing the message

Hard conversations are where managers most often soften the point until it disappears. This prompt keeps the message clear and kind, and it pushes you to ask questions before you assume.

I need to give difficult feedback to R., a senior analyst. Facts: R. missed three report deadlines in six weeks, and each delay pushed a client update back by a day. R. is skilled and well liked, and I don't know the cause yet. Write a two-sentence opening that states the issue without blame or labels like "unreliable." Then give me three open questions to understand what's going on, and one sentence that sets a clear expectation for the next month. Finally, list two reactions R. might have and a calm response to each.

Use the output as notes, not a script. If you need to confirm the expectation in writing afterward, our prompts for professional emails help keep the tone steady.

If the issue could become formal, involve HR before the meeting. In the UK, formal action over poor performance is covered by the Acas Code of Practice on disciplinary and grievance procedures, which employment tribunals take into account, and a 2026 draft revision puts even more weight on trying an informal conversation first. The US has no single national equivalent, so your company's policy sets the process, and a written plan to fix performance is often called a PIP (performance improvement plan).

Who it suits: any manager facing a performance or behavior issue that needs addressing early, while it is still a conversation and not a process.

3. Draft a performance review from your evidence

The performance review prompts that work best start with a pile of evidence, not a blank page. Gather your notes, the person's goals and their results first, then paste them under this prompt.

Act as an experienced people manager. Below are my notes on a direct report's year, their three goals and what they achieved. Organize them into a draft review with four sections: key achievements, areas for growth, specific examples for each point, and suggested goals for next year. Use only the facts I give you. If a section has no evidence, say so instead of filling the gap. Write in a direct, supportive tone, around 350 words.

The "say so instead of filling the gap" line matters. Without it, AI tools tend to invent tidy achievements that sound plausible and are not true. Check every claim against your notes.

Who it suits: managers writing several reviews in one cycle who need a consistent structure.

4. Check your review for bias and vague language

Run this before any review leaves your hands. Personality labels such as "abrasive" or "not a team player" describe the writer's impression, not the person's work, and they are hard to act on.

Review this performance review draft as a fair, skeptical HR partner. Flag vague praise or criticism with no example, personality labels instead of observed behavior, any comment on age, family, health or other personal traits, and anything that contradicts the overall rating. For each flag, quote the phrase and suggest a specific, behavior-based rewrite. Do not rewrite the whole review.

Treat the output as a second pair of eyes, not a verdict. AI can miss bias too, and it can flag comments that are fair. You decide which changes are right.

Who it suits: every manager, especially anyone rating several people with different working styles.

5. Write your own self-review as a manager

Managers have to review themselves too, and many undersell the work of leading because it does not show up as personal output. This prompt frames your year around your team's results and how you grew people.

Help me draft my self-review as manager of a five-person marketing team. This year we launched two campaigns on time, I hired and onboarded one new team member, and one person earned a promotion. One goal slipped: our reporting dashboard is three months late. Draft about 250 words under three headings: team results, developing people, and what I'll do differently. Own the missed goal plainly, without excuses. Then ask me two questions that would make this review stronger.

Answer the two questions it asks, because that is where the specific, convincing detail comes from. For a fuller walkthrough aimed at individual contributors, see how to write a self-evaluation step by step.

Who it suits: managers in an annual or mid-year review cycle, and first-time managers writing their first leadership self-assessment.

6. Plan a 1:1 that isn't just a status update

Many 1:1s drift into project updates that could have been a message. AI can draft 1:1 meeting questions tailored to where someone is right now, which gives the meeting a purpose.

I have a 30-minute 1:1 with a team member who joined four months ago. They've been quieter in team meetings lately, but their work is solid. Suggest a simple agenda and eight open questions across three themes: how they feel about the work, what's blocking them, and their growth. Avoid yes-or-no questions and anything that sounds like an interrogation. Mark the two you'd ask first.

Pick two or three questions, not all eight. A 1:1 that works through a list of eight questions feels like a survey, and the person will notice.

Who it suits: managers running weekly 1:1s, or ones every two weeks, who feel the meetings have gone flat.

7. Turn 1:1 notes into a clear follow-up

The value of a 1:1 often leaks away afterward because nobody wrote down who agreed to what. This prompt turns your notes into a short recap. Strip out anything personal the person shared before you paste.

Here are my notes from a 1:1 with A. Turn them into a short follow-up message for Slack or Teams: three bullets covering what we agreed, who owns each action and by when, plus one line thanking A. for raising the workload concern. Keep it under 100 words, friendly and plain.

For longer transcripts or meeting recordings, these prompts to summarize long meetings and documents go deeper.

Who it suits: managers juggling many direct reports, and remote teams that work mostly in chat.

8. Rehearse a difficult conversation with role-play

Reading a script is not the same as hearing pushback. Ask the AI to play your team member so you can practice staying calm and listening before the real meeting.

Role-play a conversation with me. You are T., a capable developer who is frustrated that a promotion went to a colleague. You're polite but defensive, and you believe your work was overlooked. I'll speak as T.'s manager. Reply as T. would, one message at a time, and don't make it easy. After eight exchanges, step out of character and tell me where I listened well and where I sounded dismissive.

Remember that the AI is guessing. Real people react in ways no model can predict, so use the rehearsal to steady your nerves and sharpen your opening, not to script the whole talk.

Who it suits: managers facing promotion decisions, reorganizations or any conversation they have been putting off.