A sovereign, on-device meeting intelligence system that captures, transcribes, and summarizes meetings without sending audio to third-party clouds.
ODW Recap turns meetings into structured action items, summaries, and decisions — all on your local machine. Its core promise: meeting recordings and transcripts never leave your device unless you explicitly choose a cloud provider.
No cloud bot silently joins your calls. No transcripts sit on a vendor's servers by default.
- On-device transcription: Uses Whisper-family models locally via whisper.cpp
- Local AI summarization: Generates summaries, action items, and decisions with local LLMs
- Pluggable providers: Choose from local models or cloud STT/LLM providers (OpenAI, Anthropic, Deepgram, etc.)
- Flexible audio input: Record system audio, use your microphone, import files, watch folders, or upload from your smartphone
- Encrypted storage: AES-256-GCM encryption with Argon2id key derivation
- Team server (optional): Go backend for multi-user sync, RBAC, audit logging, and admin dashboard
- Vault/Loop integration: Designed to feed meeting knowledge into ODW.ai suite modules
Privacy-conscious teams, regulated industries, and organizations that need sovereign control over their meeting data.
┌─────────────────────────────────────┐
│ Desktop App │
│ Tauri (Rust backend + webview UI) │
│ - Audio input & capture │
│ - Storage (SQLite + encryption) │
│ - Transcription (STT providers) │
│ - Summarization (LLM providers) │
│ - Prompt manager │
│ - Sync engine │
└──────────────┬──────────────────────┘
│ (optional)
▼
┌─────────────────────────────────────┐
│ Team Server │
│ Go + Chi + PostgreSQL + Redis │
│ - JWT auth & RBAC │
│ - Meeting sync │
│ - Admin dashboard API │
└─────────────────────────────────────┘
cd desktop-app
npm install
cd src-tauri
cargo buildcd team-server
go build -o recap-team-server ./cmd/server
./recap-team-serverSee DEVELOPMENT.md for the full build, test, and packaging instructions.
| Document | Purpose |
|---|---|
| prd.md | Product Requirements Document |
| sad.md | System Architecture Document |
| tsd.md | Technical Specification Document |
| tbk.md | Task Breakdown Document |
| DEVELOPMENT.md | Current implementation status and build instructions |
| research.md | Market research and competitive analysis |
| CLAUDE.md | Guidance for AI assistants working on this codebase |
- Desktop app: Tauri 1.4, Rust, vanilla JS/HTML/CSS
- Desktop database: SQLite 3.44+ with FTS5
- Backend language: Rust (Tauri commands)
- Team server: Go, Chi, PostgreSQL, Redis
- Transcription: whisper.cpp, OpenAI, Deepgram, AssemblyAI, AWS, Azure, Google
- Summarization: llama.cpp, Ollama, OpenAI, Anthropic, Google, AWS, Azure
This repository is built to be extended with AI coding agents. Rather than a turnkey product, ODW Recap is a working, well-structured codebase you can clone and adapt to your own needs with an agent like Claude Code. The repo includes agent context files (e.g. CLAUDE.md) and clear architecture docs so an agent can quickly understand the structure and help you customise, integrate, and extend it. To get started: clone the repo, open it with your coding agent, point it at this README and the docs, and describe what you want to build.
This project is licensed under the MIT License - see the LICENSE file for details.