How the AI conversation works

1. Your audio is streamed through the real-time session service.
2. Speech recognition turns the current audio into text for processing.
3. The selected coaching style and relevant session context guide the language model's response.
4. Speech synthesis turns that response back into audio.

Several specialist service providers are involved in this flow. The categories of data they process are described in the privacy policy.

You can complete the in-app questionnaire or select a style yourself. VocalCalm supports CBT, DBT, ACT, Solution-Focused, Trauma-Informed, IPT, Psychodynamic, Humanistic, and Integrative coaching.

The label changes the conversation's instructions and emphasis. It is not a diagnosis and does not assign a clinical treatment plan.

Short structured insights can be retained so a later session has useful context, such as a goal, a strategy that helped, or a next step to revisit. This memory can be managed or deleted in the app.

Operational systems may also process session data for delivery, reliability, security, and debugging. Those systems are not the same as the coach's saved memory.

During a session, VocalCalm can surface region-appropriate Helpful Resources when the conversation indicates that extra support may be useful. The action can be opened without ending the session.

This is signposting—not diagnosis, emergency monitoring, crisis intervention, or a guarantee that risk will be detected. In immediate danger, contact local emergency services.

• It can misunderstand what you mean or miss important context.
• It can produce a confident-sounding response that is incomplete or wrong.
• Its remembered insight may not capture your experience accurately.
• It cannot examine you, verify facts outside the conversation, or replace professional judgement.

• Choose or change the coaching style.
• Pause, mute, breathe, or end a session.
• Open Helpful Resources during a session.
• Manage or delete saved insights.
• Contact support about your data or account.