How Legend Uses AI
Legend uses AI to turn your audio content and metadata into transcripts, summaries, tasks, recaps, and answers. To do this, we send some data to trusted AI providers.
How it’s protected
Requests go from Legend’s servers under our account. We don’t send your name, email, or account ID, so providers can’t link requests back to you or build a profile. (The content itself, like a transcript or audio metadata, can still contain names or voices.)
We only use providers that agree in writing to protect your data at least as well as our Privacy Policy, use it only to provide the service to Legend, and not train their models on it.
We have a zero data retention policy with our model providers. No sensitive information is stored on their servers after processing — data is used only for the request and then deleted.
Data is encrypted in transit and at rest.
What we send
- Audio content and metadata from your pendant or phone (to transcribe it and identify speakers)
- Small voice samples (to recognize your voice and voices you choose to add)
- Transcripts, notes, tasks, chat messages, and files you upload (to create summaries, find tasks, and answer questions)
- Your voice, live, when you talk to Legend
Who we may send it to
Depending on the feature and language, we may send this data to: OpenAI, Anthropic, Google, Microsoft, ElevenLabs, xAI, Meta, AssemblyAI, Deepgram, Soniox, Hume AI, pyannoteAI, and Picovoice.
Your choice
These services are required for Legend to work. You can withdraw consent anytime in Settings by deleting your account and data. See our Privacy Policy and Terms of Service.