Data Policy
Version 1.0 · June 2026
Three core promises
- Your audio is never stored — not by us, not by our partners
- Your data is never used to train AI models
- Your data is never sold or shared with advertisers
How your voice is processed
Silence at the beginning and end of your recording is trimmed locally on your device before anything is sent anywhere. The audio fragment is transmitted to Groq's servers (USA) for speech recognition and deleted immediately after the text is returned. If an AI processing mode is selected, the transcription text is also processed on Groq's servers for formatting and deleted after the result is returned. In "Verbatim" mode, this step is skipped entirely. The final text is inserted into your active application.
What is sent to the server
| Data | Where | Retention |
|---|---|---|
| Audio recording | Groq (USA) | Deleted immediately after transcription |
| Transcription text | Groq (USA) | Deleted immediately after formatting |
| App context | Groq (USA) | Not stored |
| Dictionary terms | Groq (USA) | Not stored |
The name of the specific application is never sent to the server — only an anonymised category.
What stays on your device only
- Personal dictionary
- Settings (hotkey, AI mode, language)
- Dictation history in MCP IDE integration
Third-party services
Groq (groq.com) — speech recognition and text formatting. Groq's privacy policy: groq.com/privacy
Contact
Data inquiries — via the contact form at iskrio.com/contact