Privacy Policy

Who we are

Lavra is a native app for macOS and iOS, developed by MokLabs (“we”, “us”). This policy explains what Lavra does and does not do with your information.

What Lavra does with your audio

When you import a file, share a voice note, or dictate, the audio is decoded and transcribed locally on your device using Apple's on-device speech recognition. Cleanup, summaries, and titles are generated on-device using Apple's foundation models. Your audio is never uploaded to us or to any third party for processing.

• Audio files you import are stored locally in the app's container on your device.
• Transcriptions, summaries, and titles are stored locally on your device.
• We have no server that receives, stores, or analyzes your audio or transcripts.

Data we collect

None. Lavra has no analytics SDKs, no advertising, no trackers, and no user accounts. We do not collect your name, email, contacts, location, usage data, or identifiers. Per Apple's App Privacy definitions, no data is collected and no data is used to track you.

iCloud sync (your own iCloud)

If you are signed in to iCloud, Lavra can sync the text of your transcriptions between your own devices using your private iCloud (Apple's CloudKit private database). This is your personal iCloud space — we cannot access it.

• Only text (transcript, summary, title, and basic metadata such as a timestamp) is synced. Your audio is never synced and never leaves the device it was created on.
• The synced text is encrypted with a key generated on your device and stored in your iCloud Keychain, so it is end-to-end encrypted. We do not hold this key and cannot read your content.
• If you are not signed in to iCloud, the app works fully and simply keeps everything local.

Microphone

Lavra requests microphone access only when you use dictation. The recorded audio is transcribed on your device and stored locally; it is not sent anywhere. You can revoke microphone access at any time in your system settings.

When the app uses the network

Lavra is built to work offline. It connects to the internet only in these limited, optional cases — and never to send your audio or transcripts to us:

• Optional high-accuracy model (Mac): the first time you choose the high-accuracy transcription engine, the app downloads public model weights from a model host (Hugging Face). This download contains no personal data.
• iCloud sync: text sync uses Apple's iCloud as described above.

Third-party content you import

When you import audio from other apps (for example a WhatsApp voice note or an exported chat), that content is provided by you and processed on your device under this policy. WhatsApp, Telegram, and similar services have their own privacy policies governing the content while it is in those apps.

Children

Lavra is not directed to children and does not knowingly collect any data from anyone.

Your control

You can delete any transcription in the app at any time, remove microphone permission, and turn off iCloud for the app in your system settings. Because we hold no data about you, there is nothing for you to request from us.

Changes to this policy

If this policy changes, we will update the date above and post the revised version on this page.

Contact

Questions about privacy? Email support@moklabs.com.