AI for new managers: running your team’s week with an assistant, and the lines you never cross
By Bodhih Training Solutions · UpdatedThe short answer
A new manager can use an AI assistant to prepare the recurring work of the week: turning vague asks into a plan, writing delegation briefs, drafting team updates, preparing 1:1s and structuring feedback. The assistant drafts and you decide. Three lines hold in every case: AI never decides a rating, a hire or a sanction, sensitive personal details stay out of it, and it is never used to watch people.
What can an AI assistant actually do for a new manager?
Most of a new manager’s week is writing and preparation that nobody trained you for. A request from above has to become a plan. A task has to become a brief someone else can act on. A meeting has to become notes with owners and dates. A year of observations has to become a fair review.
An assistant such as Claude, ChatGPT, Copilot or Gemini is good at exactly this kind of first draft. It can structure your rough notes, suggest what you have left out, offer three ways to phrase a difficult sentence and play the other person while you rehearse a conversation. It is poor at knowing your team, your history and your company’s rules. That part is your job, and it is why the draft is never the final version.
What does a manager’s week look like with an assistant?
Think of the assistant as a capable new colleague who has never met your team. The table shows where it helps across a typical week and what stays with you.
| Task | How AI helps | What you must check or decide |
|---|---|---|
| Weekly plan | Sorts your list by impact and deadline and points out clashes | The real priorities, and whether the workload is fair across people |
| Delegation brief | Turns your notes into outcome, deadline, constraints and authority level | Whether this person is the right one, and how much support they need |
| Team or upward update | Puts the bottom line first and trims the rest | Every fact, number and name, and what is confidential |
| 1:1 preparation | Suggests questions and an agenda from your own notes | What matters to this person now. The conversation itself is yours |
| Feedback | Structures your observation into situation, behaviour and impact | That the example is accurate, specific and something you saw |
| Performance review | Drafts wording from your anonymised evidence log | The rating, the evidence behind it and any sign of bias |
| Team data | Summarises a spreadsheet and proposes a chart | At least two numbers by hand, and the recommendation |
How do you brief an assistant the way you would delegate to a person?
Weak output usually comes from a weak brief. A request like “write an update about the delay” gives the assistant nothing to work with, in the same way it would give a new team member nothing. A good brief to AI looks like a good delegation brief.
- Say who it is acting as and who will read the result.
- State the goal in one sentence: what should the reader know or do afterwards?
- Give the context it cannot guess: the background, the constraint, what has been tried.
- Show an example of the tone or format you want, if you have one.
- Specify the format and length: three bullets, a 100-word message, a table.
- Then iterate. Tell it what to change instead of asking for a fresh attempt.
Which lines should a manager never cross with AI?
The risk in a manager’s use of AI is different from the risk in a marketer’s or an analyst’s. Your drafts are about people, and people’s pay, careers and dignity sit behind them. These lines are worth holding even where no policy spells them out.
- AI never decides about a person. A rating, a promotion, a hire, a rejection or a warning is your decision, made on evidence you can explain.
- Sensitive personal details stay out. Health, family circumstances, caste, religion, complaints and salary details do not go into an assistant. Anonymise as “Team member A” before you paste anything.
- No surveillance. Do not use AI to monitor messages, score activity or infer who is disengaged.
- No invented facts. If the assistant supplies a number, a quote or an incident you did not give it, remove it or verify it.
- Some matters are not yours to draft alone. A harassment concern or a grievance goes to HR or your Internal Committee, following your company’s process.
- Use only tools your company has approved, and follow its data rules. This guide is general guidance, not legal advice.
How do you use AI for 1:1s, feedback and reviews without sounding like a machine?
Start from your own evidence, not from a blank prompt. Keep a short running log through the year: what the person did, when, and what it led to. With that log anonymised, an assistant can help you find patterns, draft specific feedback and notice where your wording is vague or harsher for one person than another.
For a hard conversation, rehearsal is the most useful and least risky use. Describe the situation without names, ask the assistant to respond as the other person might, and practise your opening line and your reply to the two reactions you fear most. Then put the transcript away. In the room, you listen.
Always rewrite the final text in your own words. If a team member would be surprised to learn how a message was written, treat that as a sign it needs more of you in it.
How do you bring your team along?
Your team is probably using AI already, each person differently. A one-page team AI charter turns private habits into shared practice. Write it with the team, not for them.
- What we use AI for, and which approved tools we use.
- What never goes in: customer data, personal details, anything confidential.
- Who checks output before it leaves the team, and for which kinds of work.
- How we say so when AI helped with something important.
- Where the shared prompts and reference documents live, and who keeps them current.
Where should a new manager start this week?
Pick one recurring task, not five. The weekly team update is a good first choice because it is low risk, it repeats, and you can compare this week’s version with last week’s. Write a proper brief, edit the draft, check every fact and send it under your own name.
Next week, add a delegation brief. The week after, prepare one 1:1 with it. By the end of a month you will know where an assistant saves you time and where it does not, and that judgement is the real skill. If you want that practice structured, graded and certified, AI for Managers is the Bodhih programme built around it.
Turn this guide into graded work on your own team
AI for Managers is Bodhih’s self-paced certification for first-time and new people managers. In about 16 hours you build eight work products on your own team, have them graded by a human reviewer, sit a proctored exam and earn a verifiable credential. It costs ₹7,999 plus GST, and access opens the moment payment is confirmed.
Questions people ask next
Can I use AI to write performance reviews for my team?
You can use AI to draft the wording of a performance review, provided the evidence is yours, the details are anonymised and the rating is your decision. Keep an evidence log through the year, give the assistant only what is needed, check the draft for bias and vague language, and rewrite it in your own voice. AI should never choose the rating or the outcome.
Should I tell my team when I use AI?
Yes, as a general rule a manager should be open with the team about using AI. Agree in a team AI charter which work AI may help with and when it should be mentioned. Routine drafting help rarely needs a label. Anything personal, such as feedback or a review, should be written and owned by you, and you should be comfortable explaining how you prepared it.
What should a manager never put into an AI assistant?
A manager should never put sensitive personal details about team members into an AI assistant. That includes health and family information, complaints and grievances, salary details, identity documents and anything a person told you in confidence. Customer data and confidential company information also stay out unless your company has approved the tool for that use. Anonymise first, and share only what the task needs.
Can AI help me prepare for a difficult conversation?
Yes. Rehearsing a difficult conversation is one of the most useful ways for a manager to use AI. Describe the situation without names, ask the assistant to respond as the other person might, and practise your opening and your replies. It helps you find clear, calm wording. It does not replace listening in the real conversation, and serious matters still go through HR.
Is it acceptable to use AI to screen job candidates?
AI can help a hiring manager with the preparation around screening, such as writing the job description, the interview questions and an anchored scorecard. It should not make the decision to reject or shortlist a person on its own. A human should review every candidate against the same criteria, candidate data should be handled with care, and your company’s hiring policy applies. This is general guidance, not legal advice.