Privacy & Security
Understand the local processing boundary for meeting audio, transcripts, summaries, screenshots, search, embeddings, and Q&A.
Step-by-Step Guide
- Meeting audio and derived meeting content are processed locally by the desktop application.
- Speech recognition, summary generation, semantic embeddings, and meeting Q&A do not require a third-party inference API.
- The meeting knowledge index is stored locally with the meeting library.
- Network access may still be used for explicit model downloads and application updates.
- Always follow the recording-consent, workplace, contractual, and legal requirements that apply to your meetings.
Learn the practice
- What a Private Local AI Meeting Assistant Should Keep On Device — Privacy is more than local recording. Transcription, summaries, screenshots, embeddings, search, and question answering all affect where meeting information travels.
Privacy Architecture: Meeting-content inference is designed to run locally. Model downloads and application updates are separate from sending meeting audio, transcripts, summaries, screenshots, or retrieval context to a hosted inference API.