Quick backstory: After spotting the buzz around Cluely (that $15M stealth AI startup), I wanted to build an open-source version that's lighter, more private, and fully customizable. No bloat, no data leaks—just efficient AI assistance right on your desktop.

This is my first dive into Tauri, and it's been a game-changer. I tried Electron first, but it felt too heavy with all the browser cruft—high RAM usage, slow starts, not ideal for something that needs to be instant and unobtrusive. Tauri + Rust keeps it snappy, native, and tiny.

Pluely is a ~10MB desktop app (27x smaller than Cluely's ~270MB) that gives real-time AI help during calls, interviews, or chats—undetectable overlay, no meeting transcription (though you could extend it).

Fresh in v0.1.1: Custom LLM support! Add any provider, including local ones like Ollama or custom URLs/APIs for full offline mode. Plus, better window management (focus/resize), improved markdown rendering, auto-scrolling chats, and layout tweaks for smoother UX.

Core features: Undetectable overlay(Tested in MacOs): Translucent window floats over apps like Zoom/Teams, hidden in shares/recordings/screenshots. Now with better click-through and positioning. Multi-LLM integration: OpenAI, Claude, Gemini, Grok with dynamic model fetching. Plus the new custom option for locals. Inputs: Text (markdown), voice (Whisper STT + auto-detection VAD), images/screenshots for analysis. Privacy/performance: 100% local processing (API keys stored in localStorage, no telemetry), 50% less CPU/RAM, <100ms startup. Offline-first except for your LLM calls. History: Local, with conversations list. Cross-platform: .dmg (macOS), .msi (Windows), .deb (Linux).

Great for discreet tips in interviews, quick code debugs in meetings, or strategy in negotiations. Free, tweakable, transparent—unlike Cluely's closed setup.

Repo: https://github.com/iamsrikanthnani/pluely

Feedback welcome! Thoughts on custom LLMs, voice tweaks, or what else for a stealth AI companion?