Processing and privacy
What stays on your device?
Convoxa has two modes. On Device transcribes, labels speakers, and writes notes on your iPhone, iPad, or Mac with Apple Intelligence. Cloud does all three with Convoxa AI. iCloud sync is a separate setting that copies your library to your personal Apple account.
Processing and sync do different jobs
| Feature | What happens | Connection |
|---|---|---|
| On Device | Transcribes, labels speakers, and writes notes on your device. Recording content is not uploaded for processing. Requires Apple Intelligence. | Works offline once the required language assets and speaker label model are installed. |
| Cloud | Uploads audio to transcribe, label speakers, and write notes with Convoxa AI, with a broader language list. | Internet required. |
| iCloud sync | Syncs recordings, transcripts, and notes to your personal iCloud account. Included on every plan. | Internet required to sync between devices. |
Prepare for offline use
Complete account setup with Sign in with Apple while online. Choose On Device, make sure Apple Intelligence is turned on, and check that your language is available on the device you will use. Required speech assets may need to download before the first offline session.
Speaker labels use an open-source model of about 100 MB, which downloads from Hugging Face automatically when you use On Device on a suitable connection. Make a short test recording before relying on the setup for a meeting.
Keep recording content only on this device
Use On Device, turn off iCloud sync in Convoxa Settings on the device, and do not share the recording or enhance the note with Cloud. On Device by itself does not disable sync: iCloud sync is enabled by default, including on the free plan.
This distinction concerns your recording content. Convoxa still uses an account for authentication and quotas; account and service metadata are described in the Privacy Policy.
What goes to the cloud in Cloud mode?
Audio is uploaded for transcription, and your transcript and any included context, such as typed thoughts and images, are processed to write the note. Uploaded audio is deleted when transcription finishes. Temporary processing inputs and results are retained for delivery and recovery, then deleted according to the periods in the Privacy Policy. Content is not used to train AI models.
If a recording does not fit in your remaining Cloud hours, it may be transcribed on your device instead. Note generation then sends the transcript and included context, but not the audio.
Which mode fits your meeting?
Choose On Device when you want processing to stay on your device and it supports Apple Intelligence. Choose Cloud when you want the highest accuracy or need its broader language coverage. Decide separately whether to sync your library.
See the language guide and plan limits before choosing your setup.