Mirror
Practice difficult conversations before the stakes are real.
THE PRODUCT
Training can explain how to approach a difficult conversation, but it cannot recreate the pressure of responding in the moment. Mirror gives managers and People teams a private voice rehearsal room with a character that listens, reacts and changes the conversation in response to what the learner chooses to say. Each practice session is built around a role, a situation and the company's criteria: giving feedback, communicating a decision, interviewing or resolving conflict. Afterwards, Mirror returns to the specific moments that changed the conversation and explains what to try differently. The learner can repeat the scenario and practice another response before the consequences are real.
IN USE
- (01)
The company defines the situation
The scenario combines a role, an objective, points of tension and the criteria to practice.
- (02)
The conversation responds
The character reacts turn by turn to the learner's questions, silences and decisions instead of following a fixed script.
- (03)
The learner reviews and repeats
Mirror points to specific moments, explains the feedback and allows another response to the same challenge.
ENGINEERING
Mirror separates the real-time simulation runtime from post-session analysis. Each scenario defines a role, objective, constraints, tension points and rubric; the runtime compiles that definition with live conversation state to keep the character, available information and objective consistent across turns. The voice layer manages turn detection and interruption, while the transcript retains timestamps and the scenario events triggered during the session.
Afterwards, the evaluation engine maps specific transcript segments to rubric criteria. Every observation retains the turns that support it, preventing generic feedback without evidence, and attempts on the same scenario can be compared on a common basis. Access, retention and identity or LMS integrations are configured per company. The evaluation suite tests persona drift, scenario adherence, feedback grounding, scoring consistency across paraphrases and conversational latency; it does not infer personality, potential or employability.
