Security & Privacy

How Prismical keeps your meetings and notes private with local-first processing and open-source transparency.

Local-First by Design

Prismical processes audio on your device by default. Your conversations stay on your machine:

Audio never leaves your device — When using local models (Whisper, Parakeet), audio is processed in memory and never uploaded
No bot joins your meetings — Prismical captures system audio silently, without joining as a participant
Local storage — Notes, transcripts, and summaries are stored on your machine. No cloud sync unless you choose it.
No account required — Use Prismical without signing up for anything

BYOK Cloud Providers

If you choose to use cloud transcription (Deepgram, OpenAI, Assembly AI):

Audio is sent to the provider you select, using your own API key
Prismical never intermediates — your key, your account, your provider's privacy policy
You can switch back to local models at any time

Open Source

Prismical is fully open source under the MIT license:

Every line of code is public and auditable on GitHub
No hidden data collection or telemetry
Community-reviewed security
Contributions and security audits welcome

No Video Recording

Prismical only processes audio for transcription. No video is captured, recorded, or stored.

While Prismical doesn't join meetings as a visible bot, you should still inform participants that the meeting is being transcribed. Many jurisdictions require consent for audio recording. Check your local laws and company policies.

Data You Control

All data is stored locally on your machine
Delete any meeting, note, or transcript at any time
No data retention policies imposed by Prismical
Export your data in standard formats (JSON, Markdown)