// the find
Zackriya-Solutions/meetily
Privacy first, AI meeting assistant with 4x faster Parakeet/Whisper live transcription, speaker diarization, and Ollama summarization built on Rust. 100% local processing. no cloud required. Meetily (Meetly Ai - https://meetily.ai) is the #1 Self-hosted, Open-source Ai meeting note taker for macOS & Windows.
Meetily is a local-first meeting transcription and summarization tool built with Tauri (Rust backend, Next.js frontend). It runs Whisper or NVIDIA Parakeet for real-time transcription and pipes results to Ollama or other LLM providers for summaries. Aimed at privacy-conscious professionals and enterprises who won't send meeting audio to cloud services.
- Genuine local processing architecture using Tauri with SQLite migrations for persistence — no hidden cloud calls for the core transcription pipeline, which matches the privacy pitch
- Support for both Whisper and Parakeet (ONNX) transcription engines with GPU acceleration paths for Metal, CUDA, and Vulkan, giving users real hardware options rather than just CPU fallback
- Modular LLM provider support (Ollama, Claude, Groq, OpenRouter, custom OpenAI-compatible endpoint) implemented as separate Rust modules, so swapping providers is straightforward
- Active migration history and a real database schema with repositories pattern suggest the data layer is taken seriously, not just a prototype SQLite dump
- Codebase shows significant accumulation of dead code: `core-old.rs`, `recording_commands.rs.backup`, `lib_old_complex.rs`, `audio_v2` alongside `audio` — this is a maintenance liability and suggests rewrites that weren't fully cleaned up
- The backend directory contains a full Python/FastAPI service with its own Dockerfiles and whisper.cpp submodule, while the README describes it as a single Tauri app — the architecture is unclear and potentially misleading to new contributors
- Speaker diarization is listed in the description and some migration columns exist (`add_speaker_field`) but the PRO page marks it 'Coming Soon', so the community edition likely ships with incomplete or non-functional diarization
- Analytics module in the Rust source (`src/analytics/`) with no clear documentation of what's collected contradicts the 100% local/privacy-first branding and will raise immediate red flags for the target enterprise audience