Privacy
Cuvo is designed to keep your audio private while still delivering powerful coaching cues. Here’s exactly what stays on your device and what reaches the cloud.
Audio: always on-device
Your audio never leaves your Mac, iPhone, or iPad. Cuvo uses Apple SpeechAnalyzer, which runs entirely on-device, to convert speech to text. No audio data is transmitted to any server — not to Cuvo, not to Apple, not to anyone.
This applies to both microphone audio (your voice) and system audio capture (the other side of a call, on macOS).
Transcript text: sent to the cloud for cues
The text transcript — the words that SpeechAnalyzer produces from your audio — is sent to a cloud-hosted AI model to generate coaching cues. The AI needs the conversation context to deliver relevant, timely suggestions.
This means:
- Your raw audio stays local.
- The text representation of what was said is processed in the cloud.
- Coaching cues are generated from that text and sent back to your device.
Config: fetched from the cloud
Cuvo fetches its configuration from the cloud at launch — conversation types, coaching parameters, and prompt templates. This is lightweight metadata, not conversation data.
Private Mode: fully on-device
When you enable Private Mode, coaching cues are generated entirely on-device using Apple Intelligence. In this mode:
- No transcript text is sent to the cloud.
- No conversation data of any kind leaves your device.
- Cues may be simpler than cloud-generated ones, since the on-device model is smaller.
To toggle Private Mode, tap the provider indicator on the setup screen (the small label that shows “Cloud” or “On-device”). Private Mode is available whenever Apple Intelligence is supported on your device.
Summary
| Data | Where it’s processed | Cloud involved? |
|---|---|---|
| Audio (mic and system) | On-device (Apple SpeechAnalyzer) | Never |
| Transcript text | Cloud AI model | Yes (unless Private Mode) |
| Coaching cues | Cloud AI model | Yes (unless Private Mode) |
| Config (types, prompts) | Fetched from cloud | Yes (metadata only) |
| Reference documents | Text extracted on-device, sent with transcript | Yes (unless Private Mode) |
| CRM contact context | Fetched from CRM, sent with transcript | Yes |
What this means in practice
Even without Private Mode, your raw audio never leaves your device. The privacy boundary sits between audio and text: Apple SpeechAnalyzer converts speech to text locally, and only the resulting text is processed in the cloud to generate coaching cues.
If you want nothing to leave your device, enable Private Mode. You’ll still get coaching cues — they’ll just come from the on-device model instead of the cloud.
For the full legal details, see our Privacy Policy.