You are talking to a machine, and here is exactly what it does.

• Writes the practice situations and the replies of the person you are talking to.
• Reads your written reply and suggests a calmer or clearer version of it.
• Rates how well the emotion word you picked fits the situation you described.
• Reads text aloud with a synthetic voice. Every voice in the app is generated, never a recording of a real person.
• Translates the interface and the practice text into your chosen language.

Anywhere you see the AI label in the app, a language model produced that text or audio.

• It never tries to detect how you feel. You choose your own emotion word and mark where you feel it in your body. Nothing is inferred from your face, your voice or your typing.
• It never makes a decision about you. No hiring, promotion, pay, ranking or disciplinary use, and no result is sent to an employer.
• It never profiles you for advertising, and your practice text is not used to train anyone else's models.
• It never watches you in the background. Nothing is recorded unless you type or tap it.

• Every body check-in and every emotion question can be skipped, and the practice still works.
• Scores, streaks and levels are yours. You can delete your data at any time and the account goes with it.
• Feedback is a suggestion, not an assessment. You decide whether it fits your situation.
• If you disagree with something the AI said, tell us and a person on our team will look at it.

Language models invent plausible-sounding detail. Coaching feedback can miss context you never wrote down, misread cultural norms, or sound more certain than it should. Treat it as a sparring partner with blind spots.

Win-Win is not therapy, medical advice, legal advice or an HR process. If a conversation involves harassment, safety or your health, speak to a qualified person. If you are in crisis, contact your local emergency service or a helpline.

• Interaction disclosure: you are told up front and in every session that the conversation partner, the feedback and the voice are AI-generated.
• No emotion recognition: the app does not infer emotions from biometric data, so it stays clear of the prohibition on emotion inference in workplace and education settings.
• Not an employment decision system: results are private to the person practising, so the app is not used to evaluate, select or manage employees.
• Synthetic media labelling: generated text, audio and avatar images are marked as generated inside the app.
• AI literacy: this page, the science section and the in-app explanations are written so any user can understand what the system can and cannot do.

We classify Win-Win as a limited-risk system with transparency duties. This is our own assessment, kept up to date as the product changes, and not a certification.

If an AI reply was wrong, unfair or upsetting, write to us through the contact form and describe what you saw. We answer within five working days, review the case with a human, and adjust the system when the complaint is fair.

• Trust, privacy & compliance
• Privacy policy
• The science behind the feedback