iamsrikanthnani/pluely

The Open Source Alternative to Cluely - A lightning-fast, privacy-first AI assistant that works seamlessly during meetings, interviews, and conversations without anyone knowing. Built with Tauri for native performance, just 10MB. Completely undetectable in video calls, screen shares, and recordings.

What it solves

Pluely is designed for high-stakes situations where switching to a browser tab for AI assistance would be disruptive or obvious, such as interviews, sales calls, live debugging, and lectures. It provides an invisible AI overlay that allows users to get instant answers and transcripts without leaving their current application or appearing in screen shares.

How it works

Pluely operates as a translucent desktop overlay with two primary modes:

  • Ask Mode: Users can ask questions via text, push-to-talk, or by capturing screenshots of their screen. It uses OCR for documents and supports attaching files for grounded answers.
  • Listen Mode: The app transcribes microphone and system audio in real-time with speaker labels. It can automatically generate suggested answers based on the conversation flow or provide them upon request.

The system is built with Tauri and React, storing data locally in a SQLite database for privacy. It supports a wide range of hosted models (GPT, Gemini, Claude, Llama) or the use of personal API keys and local AI CLIs (like Ollama).

Who it’s for

Professionals, students, and candidates who need real-time AI support during live interactions, meetings, and technical tasks without interrupting their workflow or being detected by screen recording software.

Highlights

  • Stealth Mode: Invisible to screen shares, recordings, and screenshots; can also be hidden from the Dock and taskbar.
  • Real-time Transcription: Live audio transcription with speaker labels and automatic response suggestions.
  • Flexible Model Support: Access to 200+ hosted models or integration with personal API keys and local AI CLIs.
  • Privacy-First: Local SQLite database for chats, meetings, and transcripts; conversations are not used for training.
  • Keyboard-First: Global hotkeys for summoning the overlay, capturing the screen, and managing transcripts.

Related

  • Project
    larashero3-dotcom/writing-dna-skill

    A framework for distilling an author's writing style into a structured "Writing DNA" that AI agents can use to stably and accurately replicate that style.

  • Project
    huggingface/cookbook

    A community-driven collection of practical Jupyter notebook tutorials for building AI applications using open-source tools and models.

  • Project
    icereed/paperless-gpt

    paperless‑gpt is an open‑source Docker service that plugs into paperless‑ngx, using LLMs (OpenAI, Ollama, etc.) or external OCR APIs to extract text, auto‑generate titles/tags/custom fields, and link related documents via reference numbers, producing searchable PDFs and a simple review UI.

  • Project
    chrisryugj/korean-law-mcp

    A Model Context Protocol (MCP) server that connects AI assistants to official South Korean legal databases to provide accurate, verified legal research and legislative analysis.