ai-notetaker
Open-source, local-first AI meeting notes app with native desktop helper and optional self-hosted history webapp.
Ficha actualizada 23 sept 2026, 2:57 · upstream 23 sept 2026, 2:41 · release sin release detectada
Qué es ai-notetaker
Descripción editorial construida a partir del repositorio, su documentación y sus señales de mantenimiento.
AI Notetaker records microphone and meeting audio, converts it into a transcript, and produces a summary with action items. It is open source, local-first, and uses your own AI provider keys without requiring an account or subscription. The app runs entirely on your machine using Rust/Tauri for the desktop helper and Chrome Native Messaging for the extension.
The normal user flow involves installing the desktop helper, loading the Chrome extension, completing a four-step setup wizard that checks audio devices and provider keys, then recording meetings with real-time transcript updates. After each meeting, you can open the summary page to review notes, listen to transcripts, and see action items generated by AI.
The production distribution model separates concerns: native helper installers handle capture and local storage, while Docker is reserved for the optional self-hosted history webapp that provides authenticated cross-device access. The extension communicates with the helper through Chrome Native Messaging and a Unix socket or Windows named pipe, never opening its own TCP server.
Key features include live transcript display during recording, meeting summaries after completion, customizable meeting modes and vocabulary, crash recovery via retry queues for interrupted provider calls, and an optional self-hosted webapp for authenticated history. Audio data is stored locally before being sent to providers, ensuring privacy even if the app crashes or is terminated.
The project uses Rust as its core framework with Tauri for the desktop UI, PostgreSQL for local storage, and supports multiple AI provider tiers including Deepgram + Claude, Groq + Gemini, and DeepSeek. The architecture separates concerns between audio capture, transcription, summarization, and action item generation, with retry logic handling interruptions gracefully.
Contributors can build from source using cargo commands in the helper and extension directories, while the webapp requires Prisma migrations and PostgreSQL setup. Documentation covers getting started, troubleshooting, data handling, testing strategies, and architecture decisions documented in separate guides.
- Record meetings with colleagues or team members to capture discussion points and action items
- Create meeting summaries after a session ends for quick review without reading full transcripts
- Capture brainstorming sessions or workshops where multiple speakers are present
- Store personal notes from video calls or conference recordings in one place
- Generate structured action items directly from AI-generated task lists during meetings
- Maintain meeting history across devices with optional authenticated access via self-hosted webapp
- Fully local-first architecture with no account, subscription, or backend required for basic use
- Native desktop helper written in Rust/Tauri provides a smooth tray integration experience
- Chrome extension uses Chrome Native Messaging to communicate with the helper without exposing internal APIs
- Supports multiple AI provider tiers including Deepgram + Claude, Groq + Gemini, and DeepSeek
- Includes retry logic for interrupted provider calls that preserves raw audio data locally
- Self-hosted webapp option allows authenticated cross-device history storage
- Audio recording requires a virtual-audio setup (BlackHole on macOS, VB-CABLE on Windows, PulseAudio/PipeWire null sink on Linux)
- Provider costs are not bundled and must be paid directly to the chosen AI service
- Recording permission may be required by some meeting apps or local laws depending on jurisdiction
- The extension does not open a TCP/localhost server; communication is via Native Messaging or Unix socket
- No built-in audio quality control beyond basic recording; raw audio must be processed externally
- Self-hosted webapp requires additional infrastructure and authentication setup for cross-device use
Perfil de infraestructura
Estimación inicial derivada del repositorio y de los servicios detectados.
Despliegue detectado
Qué señales reales de instalación hemos encontrado.
Install the Rust helper package (not just cargo build) to register with Chrome Native Messaging, or use a pre-built installer,Load the extension from `extension/dist/` in Chrome's Developer Mode; do not replace the key in manifest.json,For self-hosted webapp: deploy on Railway or similar platform and paste the URL into extension Settings > Web App,Use Docker for the optional history webapp to isolate it from the main application process,Configure PostgreSQL database with appropriate permissions; use environment variables for credentials,Test audio routing by selecting AI Notetaker devices in your meeting app before recording,Keep helper and extension processes running simultaneously; the extension waits for helper readiness state
Stack y dependencias
Componentes inferidos a partir de manifests, configuración y documentación.
Puede sustituir a
Relaciones detectadas por el análisis del repositorio y su categoría.
Experiencia real de la comunidad
Ayuda a contrastar la estimación automática con instalaciones reales.
Todavía no hay datos comunitarios para este proyecto.