Supported OS
- macOS: Apple Silicon (M-Series) & Intel
- Windows: x64 & ARM64
- Note: Linux (x64/ARM64) is currently unsupported.
Hardware Requirements
- RAM: Minimum 16GB.
- GPU Acceleration: Supported on Apple Silicon M-Series CPUs and NVIDIA RTX GPUs. Other hardware defaults to CPU execution.
Core Capabilities
- Record and transcribe audio/video.
- Upload and transcribe existing media (audio, video, podcasts, YouTube links).
- Identify speakers in transcripts.
- Generate customizable LLM summaries and agentic follow-up actions (integrates with MCP/agent skills).
- Semantic search across meeting transcripts.
- Direct chat interface queryable against transcripts.
Data & Privacy Model
- Transcription: Performed locally via a downloaded Whisper-class model (
nvidia/parakeet-tdt-0.6b-v3), cached on first run.
- Summarization/Chat:
- Local LLM: 100% local processing.
- Cloud LLM: Transcripts are sent to the configured external API provider.