Press Kit
Last updated: 8 July 2026 · Press contact: [email protected]
The one-line pitch: Sopravo turns any conversation into a transcript, speaker-labeled notes, and an AI summary — entirely on your iPhone. Audio, transcripts and summaries never leave the device. No cloud, no bots, no account.
Fact sheet
| What | On-device AI meeting and conversation recorder for iPhone |
|---|---|
| Platform | iOS 26+. AI summaries require iPhone 15 Pro or newer (Apple Intelligence hardware). Recording, transcription and speaker separation work on any iPhone running iOS 26. |
| Languages | App interface in 16 languages. Summaries can be written in any of those 16 languages independently of the language spoken in the recording. |
| Pricing | 7-day free trial, then weekly/monthly/yearly subscription, or a one-time lifetime purchase (tiered, first 500 buyers only, live counter). No free tier. |
| Availability | Public launch: Monday, August 10, 2026 |
| Developer | Berkay Turk, trading as BRT APPS — an independent developer based in France |
| Website | sopravo.app |
The privacy summary
Recording, transcription, speaker separation (diarization) and AI summarization all run on the iPhone. There is no Sopravo server that receives meeting content, no account, and no meeting bot. The app works fully in airplane mode, including the AI summary. Optional backup uses the user's own private iCloud database — Apple's architecture keeps it inaccessible to the developer. The only network traffic Sopravo can generate is optional, content-free diagnostics (togglable in Settings) and App Store billing. The complete, testable inventory — every network request the app can make, and three ways to verify it yourself (airplane mode, an HTTPS proxy, the analytics opt-out) — is published at sopravo.app/verify.
How it works (technical)
Transcription runs on Apple's SpeechAnalyzer framework. Summaries — executive summary, decisions and action items — are generated on-device by Apple's Foundation Models (Apple Intelligence). Speaker separation uses a local diarization pipeline: Silero VAD for voice activity detection, feeding a pyannote community-1 + WeSpeaker pipeline, entirely on-device — no NVIDIA Sortformer or other server-oriented models. This paragraph is safe to quote verbatim.
Screenshots & assets
Device-captured screenshots (light + dark, English and German) and the app icon are being finalized ahead of the August launch. Email [email protected] for the current asset zip or a TestFlight build — happy to send either well ahead of the embargo.
Embargo
Any information shared ahead of public launch is embargoed until Monday, August 10, 2026, 15:00 CEST / 9:00 AM ET, unless otherwise agreed in writing.
Contact
Press inquiries: [email protected] · General support: [email protected]