Solo practice
Rehearse with an AI employee
Any time, as often as needed, with no one watching. The AI reacts to what the manager says, so no two runs are the same.
Difficult conversations training with AI roleplayThe conversation you keep putting off. Practice it first.
Managers practice feedback, underperformance and bad-news conversations with an AI employee who pushes back, deflects or gets upset. After each run they see what to change, and run it again.
AI characterVoice or video
Finn · AI senior engineer
“Everyone on the team is behind. Why am I the one in this meeting?”
AI character’s mood1:32
- Opens with the facts
- Acknowledges the reaction
- Stays on the purpose
- Agrees on a next step
AI coachDon’t argue about the team. Bring it back to his two deadlines.
Why practice
Postponed conversations don’t go away. They get harder.
Most managers know which conversation they are avoiding. What they lack is a safe place to get the first attempt wrong. AI roleplay gives them one, as many times as it takes.
AI and expert assessorsr = 0.80
correlation between AI scores and expert scores on the same conversations
Per run5–20 min
short enough to rehearse the night before the real talk
Languages50+
practice in the language the conversation will happen in
Conversation examples
The talks managers put off. Each one with a character who resists.
Every scenario has an AI character with a backstory, a reason to push back and criteria from your competency model. Your own situations can be added in minutes.
- Underperformance
NiaAI account leadA target missed for the second quarter
“I’m doing everything I can. The market just isn’t there.”
- Behavior feedback
LucasAI data analystInterrupting colleagues in every meeting
“I’m passionate about the work. Is that a problem now?”
- Declined promotion
MeiAI operations analystTelling her she didn’t get the role
“What did the other candidate have that I don’t?”
- Team conflict
OmarAI shift supervisorTwo team leads who stopped talking
“Ask him why he went to your boss instead of me.”
- Restructuring
DavidAI regional managerA role that won’t exist after the reorganization
“How long have you known about this?”
- Personal situation
SofiaAI customer success leadPerformance that dropped after a hard few months
“I didn’t think it was showing at work.”
- TerminationSamAI voice
Ending employment, with dignity
“Is there anything I could have done differently?”
- Customers
Angry customers and complaints
Frontline and service teams rehearse escalations, refunds and de-escalation with AI customers.
Customer service training →
What gets scored
Four skills hold a hard conversation together. Each one is scored.
Use these as they are, or load your own competency model and the AI scores against it instead.
Facts, not labels
States the purpose in the first minute and describes behavior and its impact, not the person.
Listening first
Asks, pauses and summarizes what was said before answering an objection.
Handling the reaction
Names the other person’s reaction without judging it and stays steady when it escalates.
A next step
Ends with something concrete: what changes, who owns it and when you will talk again.
The turning point
Same moment, different words. A different conversation.
The AI employee reacts to what the manager actually says. One sentence can open the conversation or end it, and the feedback shows which one it was and why.
Finn · AI employee
Everyone on the team is behind. Why am I the one in this meeting?
Finn
Okay. The first one was the API dependency. The second one… that was me.
AI character’s mood
+12Facts, not labels
AI coachYou answered his question and narrowed the focus. He owned the second miss himself.
Want one of your real situations turned into a scenario like this?
Book a demoTwo ways to practice
Alone with the AI. Or across the table from a peer.
Peer mode
Take turns with a colleague
Two managers swap roles, one playing the employee. The AI listens to each run and scores one side or both against the same model.
For L&D teams
From one hard talk to a whole program.
Before one conversation
A manager rehearses the real situation
- Scenario built from a short description
- Two or three runs the evening before
- Private by default
A manager cohort
A program for new or newly promoted managers
- Scenarios mapped to your competency model
- Peer practice in pairs
- Progress by manager and by skill
Across the company
One standard for hard conversations
- SSO, and any LMS through SCORM or xAPI
- Reports by team and region
- 50+ languages
IT and tech · first-time managers−40%
voluntary turnover once new managers practiced their hardest conversations before having them for real.
- Challenge
- Engineers moved into management without people skills. The result was friction in their teams and high attrition.
- Program
- Rehearsal ahead of the high-stakes talks: performance reviews, salary conversations and burnout check-ins.
- Performance review
- Salary conversation
- Burnout check-in
- Giving feedback
- Delegation
Client name withheld · methodology available on request
FAQ
What L&D teams ask us. Answered plainly.
What is difficult conversations training with AI?
Managers rehearse a hard conversation with an AI character by voice or video. The character reacts to what they say, pushes back and gets upset like a real person, and every run ends with a score per skill and timestamped comments.
Which difficult conversations can managers practice?
Underperformance, behavior feedback, declined promotions and raises, team conflict, restructuring, termination, personal situations that affect work and pushing back on your own manager. Company-specific cases can be added. One is open to read in full: a conflict between two engineers, with Kenji as the AI engineer.
What makes a difficult conversation go well?
Preparing the facts and the purpose, stating them early, listening before answering, naming the other person’s reaction without judging it and ending with a concrete next step. Scenarios score each of these.
Can managers rehearse a conversation with their own boss?
Yes. Upward conversations, such as pushing back on a deadline or asking for resources, are scenarios like any other, with an AI senior leader who has their own priorities.
How realistic is the AI character?
The character reacts to the manager’s words: it can cooperate, deflect, go quiet or escalate. How hard it pushes back is set per scenario, so managers can start calm and work up to hostile.
Can we add our own situations?
Yes. Describe a real case and the scenario builder sets the character’s role, backstory and criteria. Managers can also rehearse a specific upcoming conversation privately.
Does the AI analyze the manager’s emotions?
No. It scores what the manager says and how the conversation is run, so skills like empathy are judged by the content of each reply. Emotional states are modeled only for the AI character, as part of the scenario.
Who sees the results?
It is configurable: a private practice space, a view for L&D and HR, or both at different stages of a program.
How is this different from a workshop?
A workshop teaches a model; practice makes it usable under pressure. Most teams run both: the workshop first, then repeated AI practice on the situations their managers face.
The talk you’ve been avoiding. Rehearsed this week.
Book a short demo. Bring one real situation, and we will run it as a scenario on the call.