Using AI to write a performance review means giving an AI chatbot your raw notes - specific examples, ratings, and observations - and asking it to shape those into a clear, well-structured draft. The AI does the writing; you supply every fact and judgment in it, because every fact and judgment in a good review must come from the manager, not the machine.
This is one of the most practical ways to use AI in everyday life as a manager. It does not replace your thinking - it removes the blank-page problem that makes review season so draining.
TL;DR
- You can use an AI chatbot - such as ChatGPT (OpenAI's AI chatbot) or Google Gemini (Google's AI assistant) - to turn your raw notes about an employee into a polished, structured performance review draft.
- AI handles the writing; you supply every fact, example, and judgment. The review is only as honest and accurate as the notes you bring to it.
- The process has four steps: gather your notes, choose an AI tool, write a strong prompt, then review and edit the output carefully before it goes anywhere official.
- Never paste sensitive personal data into a public AI tool without checking your company's data policy first.
- AI-drafted reviews still need a human sign-off - treat the output as a first draft, not a finished document.
Why Performance Reviews Are So Hard to Write
Most managers are not bad at assessing their team - they are bad at translating a year's worth of mental notes into coherent, fair prose under time pressure. The result is reviews that are either vague ("a great team player") or uncomfortably blunt with no constructive framing.
AI solves exactly that problem. It takes the raw material you already have in your head - or in a quick bullet list - and gives it shape and language. The thinking is still yours.
Step 1 - Gather Your Notes Before You Touch Any AI Tool
This step has nothing to do with AI, and it is the most important one. If you skip it, the AI will produce a generic, hollow review that could apply to anyone.
Before you open a chatbot, write down:
- Two or three specific examples of work this person did well, with enough context to explain why it mattered.
- One or two areas where they fell short, described in terms of behaviour or output, not personality.
- Any goals they were set at the start of the review period, and whether they met them.
- A rough overall rating or impression - strong performer, meeting expectations, needs improvement, etc.
- Anything forward-looking - a development goal, a promotion consideration, a skill to build.
These notes do not need to be polished. Bullet points, half-sentences, and shorthand are fine. The AI will do the polishing.
Step 2 - Choose an AI Tool
For most managers, a general-purpose AI chatbot is all you need. ChatGPT (OpenAI's AI chatbot), Google Gemini (Google's AI assistant), or a similar general-purpose AI chatbot can all handle this task well.
Most general-purpose AI chatbots offer both free and paid plans; features, context limits, and data handling vary - check the provider's current plan details before using any of them.
One important caution: most consumer-facing AI tools send your input to external servers. Do not paste full legal names, employee ID numbers, salary figures, medical information, or any other personally identifiable details into a public tool. Use a job title or a first name only, and keep examples behavioural. If your company has an enterprise AI subscription with stronger privacy terms, use that instead.
Step 3 - Write a Strong Prompt (With Examples)
The quality of your output depends almost entirely on the quality of your prompt. A vague prompt produces a vague review. A specific prompt produces something you can actually use.
The Basic Prompt Structure
A good performance review prompt has four parts:
- Role context - tell the AI who you are and what you need.
- Employee context - job title, level, and review period (no personal identifiers).
- Your raw notes - paste in the bullet points you wrote in Step 1.
- Format instructions - tell the AI how you want the output structured.
Example Prompt
"You are helping a manager write an annual performance review. The employee is a mid-level marketing coordinator. Here are my notes:
- Led the relaunch of our email newsletter; open rates improved noticeably after the redesign.
- Missed two project deadlines in Q3 due to poor prioritisation; we discussed this and she adjusted well in Q4.
- Strong collaborator - the design team specifically praised her communication style.
- Goal for this year was to take ownership of one campaign end-to-end; she did this successfully.
- Development area for next year: building more confidence in presenting data to senior stakeholders.
Please write a structured performance review of around three to four paragraphs. Use a direct, warm, and professional tone. Organise it as: overall summary, strengths with examples, areas for development, and forward-looking goals."
That prompt gives the AI everything it needs. The output will be a genuine reflection of your assessment - not a generic template.
Ask for Variations
If the first draft does not feel right, you do not need to start over. Try follow-up prompts like:
- "Make the feedback on the missed deadlines more direct without sounding punitive."
- "The tone feels too formal - can you make it warmer and more conversational?"
- "Add a closing sentence that feels encouraging without being hollow."
Iteration is fast and free. Use it.
Step 4 - Edit, Verify, and Make It Yours
AI drafts are starting points, not finished documents. Before this review goes anywhere, do the following:
Read it aloud. If a sentence sounds like something you would never say, rewrite it. The review should sound like you.
Check every claim. The AI can only work with what you gave it. If a phrase implies something you did not actually observe, cut it. You are accountable for every word in this document.
Remove any filler praise. AI has a tendency to soften everything. If the employee genuinely needs to improve in an area, make sure the draft still says so clearly - not buried under qualifications.
Run it past HR if required. Many organisations have a review process or legal requirements around performance documentation. A well-written AI draft does not exempt you from those steps.
This is also a good moment to think about how AI can save you time across other writing tasks - the same prompt-and-edit approach works for emails, project updates, and more.
A Note on Tone for Difficult Reviews
Writing a review for someone who is struggling is often the hardest part of the job. Managers either over-soften the message until it loses meaning, or they go too blunt and create a document that feels punitive rather than constructive.
AI can help here specifically. In your prompt, ask it to:
- "Frame the feedback around specific behaviours, not character."
- "Suggest language that is honest about the performance gap while focusing on what improvement looks like."
- "Avoid euphemisms - the message needs to be clear."
The AI will not make the difficult conversation easier, but it can help you find language that is both truthful and fair. You still need to read the output carefully to make sure it reflects the genuine seriousness of the situation - AI sometimes softens even when you have asked it not to.
Building Your Confidence With AI Tools
If this is your first time using AI for a writing task, the process can feel uncertain. The most common mistake beginners make is writing a prompt that is too short and then being disappointed by a generic result. The fix is almost always to add more context.
If you want to build a more systematic understanding of how to prompt AI tools effectively - not just for reviews, but across your work - AILE, a gamified AI-learning app, teaches prompting skills through short, practical lessons designed for people who are new to these tools and want to use them with confidence rather than guesswork.
For the review task specifically, you can also look at how to use AI to write emails - the prompting principles are almost identical, and practising on lower-stakes writing first builds intuition quickly.
Common Mistakes to Avoid
- Prompting without notes. If you ask the AI to "write a performance review for a good employee," you will get a useless generic paragraph. Your notes are the raw material.
- Skipping the edit. AI output is a draft. It needs a human pass before it is a document.
- Pasting sensitive data. Use role titles and behavioural descriptions, not personal identifiers.
- Treating AI praise as your own observation. If the AI adds a compliment you cannot back up with a real example, delete it.
- Using the same prompt for every employee. Adjust the context and notes for each person. A review should reflect the individual.
Frequently Asked Questions
Is it ethical to use AI to write a performance review?
Using AI to write a performance review is ethical as long as you treat the AI as a drafting assistant, not a decision-maker. Every fact, rating, and judgment must come from your own direct observation of the employee - the AI only helps you express those observations more clearly. If the review accurately reflects your honest assessment, the drafting tool used to write it is a matter of process, not integrity.
What should I never include in an AI prompt for a performance review?
When prompting AI for a performance review, avoid including full legal names, employee ID numbers, salary details, medical information, or any other personally identifiable or sensitive data. Use a role description or first name only, keep examples behavioural rather than personal, and always check your organisation's data-handling policy before using any external AI tool.
Will my team know I used AI to write their review?
Your team will not automatically know AI was used to draft a performance review, and there is no obligation to disclose it - just as managers have always used templates, HR guides, and writing tools. What matters is that the content accurately reflects your genuine assessment. If your company has a policy on AI use in HR processes, follow it.
Can AI write a performance review for a poor performer without sounding harsh?
Yes - AI is particularly useful for writing performance reviews about struggling employees, because a well-crafted prompt can ask the tool to frame critical feedback constructively and focus on specific behaviours rather than character. Give the AI your concrete examples of the gap, then ask it to suggest language that is direct, fair, and focused on improvement. Always read the output carefully to make sure it still reflects the seriousness of the situation.
Do I need a paid AI plan to write a performance review?
Most general-purpose AI chatbots offer a free tier that is sufficient for drafting a performance review, but context limits, output quality, and data-privacy terms differ between free and paid plans. Check the provider's current plan details before you start - especially if the review is lengthy or contains sensitive workplace information.
How do I make sure the AI output sounds like me, not a robot?
To make an AI-drafted performance review sound like you, read it aloud after editing and swap any phrases that feel stiff or generic for words you would actually say. You can also include a line in your prompt such as "write in a direct, warm, and straightforward tone" - and after the first draft, ask the AI to revise any sentence that sounds overly formal. One careful editing pass is usually enough to restore your voice.
Keep going with AILE
Learning AI shouldn't feel like falling behind. AILE, a gamified AI-learning app, turns it into short, friendly, hands-on lessons you can actually finish - no jargon, no gatekeeping. Join the waitlist for early access →