Rehearse the conversations you'd rather avoid.
Honest feedback, backed by your own words. And measurable progress from one attempt to the next.
You practice with an AI, not a real person. Fictional cases only, no real personal data.
How it works
- 01
Choose a scenario
Realistic conversations from everyday work
- 02
Speak or type
With an AI counterpart, in four languages
- 03
Get evidence-based feedback
Every rating quotes your own conversation, following an open methodology
Start here
New here? These three are a good place to start.
Setting expectations with a new hire
A new hire's work is below the bar after 60 days.
Delivering tough feedback
A strong performer has missed three deadlines. Address it without losing them.
Behavioral interview: tell me about a time...
A hiring manager digs into your past behavior with follow-up questions.
All scenarios
De-escalating team conflict
Two leads are clashing in public. One comes to you angry.
Salary negotiation
You've been offered a promotion. Negotiate the compensation.
Hostile all-hands Q&A
Announcing a reorg. An employee asks a pointed question in front of everyone.
Leading a team skeptical of an external hire
You're the new outside lead. A loyal veteran is politely withholding.
Managing resistance to a new digital process
A team member secretly keeps the old spreadsheet alongside the new tool.
Negotiating an unrealistic deadline with leadership
Your director committed the team to a deadline you can't hit. You have ten minutes.
What an evaluation looks like
Fictional example for illustration, not a real person's conversation. Scenario: Giving tough feedback. Jordan is an AI.
Jordan
You wanted to talk to me? I have a client appointment soon.
You · Turn 1
Thanks for making time. I'd like to talk to you about the reports from the last few weeks.
Jordan
Oh. I thought they were fine.
You · Turn 2
Well, there were a few delays, but nothing dramatic. Maybe we can just look at how things go from here.
Jordan
Okay. But what exactly do you mean by delays? I've really had a lot on my plate.
You · Turn 3
The monthly report was three days late in September and the quarterly report two days last week. I expect reports to reach me by the fifth of the month.
Jordan
That's true, and I'm sorry. The work with the new clients got on top of me.
You · Turn 4
That sounds stressful. What exactly took up most of your time?
Jordan
Mostly the constant queries from two clients. I don't get to the reports anymore.
You · Turn 5
Good, then here's what we'll do: from October you send me the reports by the fifth, and on 15 October we'll briefly check whether it's working.
- Turn 2: „Well, there were a few delays, but nothing dramatic.“ → The core message is watered down.
- Turn 3: „I expect reports to reach me by the fifth of the month.“ → Concrete expectation with examples and a date.
- Missing for the next level: state the core message early and unambiguously.
- Turn 4: „What exactly took up most of your time?“ → Open question about the cause.
- Turn 5: „Good, then here's what we'll do: …“ → The answer about the client queries is not picked up before the solution arrives.
- Missing for the next level: mirror in your own words and check understanding.
Second attempt, same case
Overall score 60 → 70
Empathy 70 → 70 · Directness 50 → 50 · Active listening 50 → 70 · Clear next steps 70 → 90
Focus applied: yes. Turn 4: „If I understand you correctly, the queries from the two clients tie you up so much that the reports get left behind. Is that right?“
Your focus for the next attempt
Your focus: mirror in your own words what your counterpart says before suggesting a solution.
How you could have said it
„Good, then here's what we'll do: from October you send me the reports by the fifth, and on 15 October we'll briefly check whether it's working.“
„If I understand you correctly, the queries from the two clients tie you up so much that the reports get left behind. Is that right? Then let's think about how we can ease the queries and you can finish the reports by the fifth.“
Why: Mirrors the core, checks understanding and connects listening with a concrete action.
Evaluated with scoring rubric v1.5, research-informed, not independently validated. A practice aid, not a replacement for coaching. View the scoring rubric
Why not just use an AI chat?
A free-form AI chat can also play through conversations, if you know how to guide it. Ensavo takes that work off your hands and makes sure every exercise is structured the same way and checkable.
Getting started
Free AI chat
Describe the role, goal and counterpart yourself
Ensavo
19 ready-made scenarios with situation, role, goals and the counterpart's hidden context
Feedback report
Free AI chat
Format and scale depend on the prompt and can differ every time
Ensavo
Four dimensions, five fixed levels with fixed points (10–90), open scoring rubric
Evidence
Free AI chat
Quotes are possible, but not checked automatically
Ensavo
Every level quotes your conversation; the app verifies that the quotes appear. Without evidence it says “not ratable”
Repeating
Free AI chat
You decide yourself what to work on
Ensavo
One clear focus per attempt, then a comparison per dimension (“Active listening: 50 → 70”)
Help while practicing
Free AI chat
Ask for it yourself
Ensavo
“I'm stuck” help, cheat sheet as PDF
History
Free AI chat
Buried in the chat history, without structure
Ensavo
Progress page, PDF per attempt, JSON backup
The framework
Free AI chat
An account with the provider, your own privacy review
Ensavo
No account, history stays in your browser, the AI is always labelled as an AI
What we do not claim yet: The rubric is research-informed, but not independently validated. The comparison with human coaches is ongoing. See the methodology
Why Ensavo exists
In more than 20 working years I have time and again seen the same patterns: feedback that comes too late or too vaguely. Roles and responsibilities that nobody has clearly named. Tasks handed out without context or reasons. Negotiations that end without a concrete result. It often starts with a conversation that never happened, or that did not go well. That costs time, trust and good collaboration.
I am convinced that leaders can hold conversations like these more confidently and more fairly if they are able to practise beforehand. Ensavo is my attempt to make that practice accessible: in a protected setting and with honest feedback that is based on your own conversation.
Melisa Mustafovic-Brüstle, Developer of Ensavo
Are you a coach or responsible for people development? I am looking for pilot partners. Write to me
Related research
The idea of practising difficult conversations with an AI role-play character is also being studied scientifically. A research team at Stanford University developed a system with which people simulate conflict conversations and receive feedback on their conversation strategies. In a study with 40 participants, those who had practised with the system beforehand used cooperative strategies about twice as often in a subsequent real conflict conversation — and competitive strategies (e.g. power or rights strategies) on average around 67% less often — than the comparison group.
Our app was developed independently of this and works with its own scoring model. The study's results refer to the system examined there and cannot be transferred directly to our app. They do show, however, that simulated practice is a promising approach — one we are continuing to examine with our own pilot phase.
Source: Shaikh, O., Chai, V., Gelfand, M. J., Yang, D. & Bernstein, M. S. (2024). Rehearsal: Simulating Conflict to Teach Conflict Resolution. Proceedings of the CHI Conference on Human Factors in Computing Systems (CHI '24). arXiv ↗