Performance review training with AI roleplayEvery review gets a rehearsal. Before the one that counts.
Managers practice performance reviews with an AI employee who agrees, argues, goes quiet or disputes the rating. Every run is scored on your review process, so the real conversation is never the first attempt.

“Meets expectations? I shipped the migration two weeks early.”
- Opens with the purpose
- Evidence before the rating
- Hears their side
- States the rating clearly
- Agrees on goals and dates
Why rehearse a review
A review happens once a year. The practice shouldn’t.
Most managers walk into a review with a form and no rehearsal, and the first time they hear an employee dispute a rating is in the meeting itself. AI roleplay moves that moment to before the meeting, as many times as they need.

correlation between AI scores and expert scores on the same conversations
a full review rehearsal fits between two meetings
managers rehearse in the language of the real review
The review, step by step
Six moments in every review. Each one rehearsed.
The AI employee reacts to every step, from the opening line to the written summary, and each step is scored on its own criteria.

So is this the conversation where I find out my rating?
It is. I’d like to walk through the year first, then the rating, then what’s next. Does that work?

Sure. I’d rather know where I stand, though.
Fair. The short version: a strong first half and a hard second half. Let’s look at both.
- States the purpose
- Sets the agenda
- Answers the direct question
- Hands over the floor

You wrote that my communication needs work. What does that mean?
In March and in May, the night shift learned about schedule changes on the day itself.

March was the system outage. That wasn’t on me.
Agreed, March wasn’t. May is the one I want to talk about.
- Uses specific examples
- Separates facts from opinion
- Concedes a fair point
- Stays on the behavior

Honestly, the rating feels unfair. I carried the migration.
Tell me what you’d want me to weigh that I might have missed.

The weekends. Nobody saw the weekends.
That’s useful. Let’s talk about why that work stayed invisible.
- Asks before answering
- Holds off on defending
- Summarizes what they said
- Finds the real issue

Below expectations? After eleven years here?
Yes. This rating is about this year, not about the eleven before it.

Then what were all those years for?
They’re why I want this year to be the exception. Let’s plan how.
- States the rating clearly
- Keeps it about this year
- Acknowledges the reaction
- Moves to the plan

I want the team lead role next year.
Then let’s make it concrete. What would show both of us you’re ready by June?

Running quarterly planning, maybe. And mentoring the new hire.
Good. Let’s write both down, with a check-in in March.
- Lets them propose
- Links goals to the review
- Makes goals measurable
- Sets a check-in date

Is that it? Do I get to see what you write?
Yes. I’ll send the summary today, and you can add comments before it goes to HR.

Okay. That actually helps.
Before we finish: is there anything you expected me to say that I didn’t?
- Summarizes agreements
- Explains what happens next
- Confirms in writing
- Invites a last word
Hard versions
The reviews managers dread. Rehearsed until they don’t.
Each scenario has its own AI employee, a backstory and criteria from your review process. Managers repeat the one they are worried about.
Disputed ratingA senior who expected “exceeds”
“I outperformed half the team. Show me the calibration.”
- Evidence
- Calibration
- Composure
UnderperformanceA missed year with a reason behind it
“I’ve been caring for my father since spring. I didn’t want to bring it up.”
- Empathy
- Clear standard
- Support plan
Former peerReviewing someone who was your peer last year
“Last year we complained about this process together.”
- Authority
- Fairness
Pay questionA raise request in the middle of the review
“Let’s be honest. This is really about my salary band.”
- Honesty
- Budget limits
- Options
Remote employeeA review over video with someone you rarely see
“You only notice my work when something breaks.”
- Recognition
- Visibility
- Check-ins
- Your process
Bring your own form and rating scale
Upload your review template and competency model. Scenarios and scoring follow your process, not a generic one.
Talk to us about your process →
Feedback after every run
What to stop saying. And what to say instead.
After each run the AI marks the lines that shut the employee down and the ones that opened them up. Managers see their first attempt next to their best one, with a score for every review skill.
I’ll keep this short, I know you’re busy.Tells him the review doesn’t matter.
I blocked an hour, because this year deserves a proper conversation.Sets the weight of the meeting.
You’re just not much of a team player sometimes.A label, not evidence. He can only deny it.
In the October release, QA learned about the change from the logs.Specific and checkable. He engaged with it.
Look, the rating is final.Ended the discussion before he finished.
Walk me through what you think I missed.He named the real issue: invisible weekend work.
Any questions? No? Great.A closed question at the end gets silence.
What would make next year’s review an easy one for you?He left with two goals of his own.
Want your own review form and rating scale scored like this?
Book a demoFor HR and L&D
Ready before review season. Not during it.
Start a review-readiness program four weeks before your cycle opens. Managers practice on their own time, and HR sees who is ready before the first real conversation.
- Week −4
Load your process
Your review form, rating scale and competency model become the scoring criteria.
- Week −3
Assign the right reviews
Each manager gets the conversations they will actually have: a disputed rating, a missed year, a top performer.
- Week −2
Practice and repeat
Five to twenty minutes per run, as often as needed. Managers can also practice against each other.
- Week −1
See who is ready
A readiness view by team shows who needs one more run or a session with HR.
- AKAnna K.3 runs · disputed rating84Ready
- TBTom B.4 runs · underperformance88Ready
- MDMarc D.2 runs · former peer71One more run
- SMSara M.1 run · disputed rating52HR session
Result from a live program
Reviews rehearsed, not improvised. Turnover down 40%.
voluntary turnover after leadership practice that included performance review scenarios.
- Challenge
- High attrition in engineering teams run by engineers recently promoted into management, before they had built people skills.
- Program
- Managers rehearsed their hardest people conversations in advance: performance reviews, salary talks and burnout check-ins.
- Performance review
- Salary conversation
- Burnout check-in
- Giving feedback
- Delegation
Client name withheld · methodology available on request
FAQ
What HR teams ask us. Answered plainly.
What does performance review training with AI look like?
A manager opens a scenario, meets an AI employee by voice or video and runs the review from the opening to the agreed goals. A run takes 5 to 20 minutes and ends with a score for each review skill and timestamped comments. One is open to read in full: a review with Nia, an account lead who disputes her rating.
What should managers not say during a performance review?
Labels instead of evidence, such as “not a team player”, criticism the employee hears for the first time, comparisons with colleagues, and promises about pay or promotion they cannot keep. The AI flags these lines in practice, so managers hear them before an employee does.
Which questions should a manager ask in a review?
Open questions that hand the floor to the employee: what they are proudest of, what got in the way, what they need from their manager and what would make next year’s review an easy one. Scenarios score whether the manager asked and listened before answering.
Can we use our own review form and rating scale?
Yes. Upload your review template, rating scale and competency model. They become the scenario context and the scoring criteria. The same competency model can also drive 360 feedback collected in AI interviews before the review.
Does it work for mid-year and probation reviews?
Yes. Annual reviews, mid-year check-ins, quarterly conversations and end-of-probation reviews are separate scenarios with their own criteria.
Can HR see how managers did?
It is configurable: a private practice space for managers, a readiness view for HR, or both at different stages of the program.
Does the AI evaluate our employees?
No. The AI plays the employee and scores the manager’s conversation skills. It does not rate real employees or read anyone’s emotions; only the AI character’s reactions are part of the scenario.
Which languages can managers practice in?
More than 15, including English and French, so the rehearsal happens in the language of the real review.
Does it replace HR training on the review process?
No. It adds what process training leaves out: the conversation itself, repeated until it feels routine. Reviews are one part of AI roleplay for new managers, next to feedback, one-on-ones and difficult conversations.
Your next review cycle, rehearsed. Before it opens.
Book a short demo. We will build a scenario from your review template and run it with you on the call.


