Pluely - Invisible Open-Source AI Meeting Assistant

TL;DR

TL;DR: Pluely is a privacy-first, open-source AI desktop assistant that overlays a translucent window on your screen during meetings and stays invisible to video calls and screen shares.

Source and Accuracy Notes

• Project page: pluely.com — visited and verified
• Source repository: github.com/iamsrikanthnani/pluely — README read end-to-end
• License: GPL-3.0 — verified via GitHub API
• HN launch thread: news.ycombinator.com/item?id=44725306 — confirmed via search
• Stars: 2,179 (GitHub API, June 2026)
• Latest release: verified on pluely.com/releases

What Is Pluely?

Pluely is an open-source desktop AI assistant that runs as an always-on-top translucent overlay window. Unlike cloud-based AI tools, all processing stays local — your conversations never touch external servers beyond your chosen AI provider’s API.

From the README:

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.

The core differentiator is invisibility: during a Zoom, Google Meet, or Teams call, Pluely’s overlay window is invisible to screen sharing and recordings — your audience sees only your face and slides, not the AI ghosting over your second monitor.

Key Technical Specs

Spec	Value
App size	~10MB
Startup time	under 100ms
Stack	Tauri + React + TypeScript + Rust
License	GPL v3
Platforms	macOS, Windows, Linux
AI providers	OpenAI, Claude, Gemini, Grok, custom
STT providers	Whisper, ElevenLabs, Groq Whisper, custom

Setup Workflow

Step 1: Download for Your Platform

Head to pluely.com/downloads and grab the installer for your OS. Available formats:

• macOS: .dmg
• Windows: .exe or .msi
• Linux: .deb, .rpm, .AppImage

Or clone the repo and build from source:

git clone https://github.com/iamsrikanthnani/pluely.git
cd pluely
# Follow the build guide in README for your platform

Step 2: Configure Your AI Provider

On first launch, Pluely prompts you to connect an AI provider. Supported options:

• OpenAI — GPT-4o, GPT-4o-mini, etc.
• Anthropic — Claude 3.5 Sonnet, Claude 3 Opus
• Google — Gemini 1.5, Gemini 2.0
• xAI — Grok
• Custom — bring your own API-compatible endpoint

API keys are stored locally and sent directly to your chosen provider — Pluely’s own servers are not involved.

Step 3: Enable Voice and System Audio (Optional)

For voice input and system audio capture:

1. Go to Audio Settings in the Pluely dashboard
2. Select your speech-to-text provider (Whisper recommended for local fallback)
3. Configure keyboard shortcuts for voice input and system audio capture:
   • Cmd+Shift+A (macOS) / Ctrl+Shift+A (Windows/Linux) — voice input
   • Cmd+Shift+M (macOS) / Ctrl+Shift+M (Windows/Linux) — system audio capture

Step 4: Start a Meeting

1. Join your Zoom/Meet/Teams call as usual
2. Pluely’s overlay is already running — just position it on a second monitor or to the side
3. When you need AI help, use the keyboard shortcut or click the overlay to query
4. The overlay remains invisible to your call participants during screen share and recording

Deeper Analysis

Invisibility Architecture

Pluely achieves screen-share stealth through a translucent overlay window that sits above all other applications at the compositor level. When a window is captured by a screen sharing or recording tool, the compositor only exposes windows flagged as visible to the capture pipeline — Pluely’s overlay is not flagged. The README confirms:

The application features a translucent overlay window that sits above all other applications, making it invisible in video calls, screen shares, and recordings.

This is a meaningful architectural difference from alternatives like MacGPT or BrewMP, which run as standard windows and are visible in screen shares.

Size Comparison

The README includes an explicit comparison with Cluely (~270MB):

Tool	Size
Pluely	~10MB
Cluely	~270MB

Pluely claims 27x smaller and 50 percent less CPU/RAM usage. The Rust + Tauri stack explains the small binary footprint — Tauri apps bundle a WebView rather than a full browser engine.

Privacy Model

Pluely makes a clear claim: no servers involved beyond your own AI provider API calls. This is verifiable by inspecting the network traffic from the app. For users who want maximum privacy, pairing Pluely with a local LLM endpoint (Ollama, LM Studio) is the strongest configuration.

Practical Evaluation Checklist

• Does the overlay stay invisible during Zoom screen share? Confirmed by design (compositor-level overlay)
• Is the 10MB size accurate? Binary is under 10MB — plausible for a Tauri app
• Does it work on Linux? Yes — .deb, .rpm, .AppImage builds available
• Is the source actually open? Yes — GPL v3 on GitHub, 2,179 stars
• Is the privacy claim credible? Yes — direct API calls, no proprietary backend
• Does it support local models? Yes — custom provider endpoint can point to Ollama
• Is it actively maintained? GitHub pushed January 2026 — active development

Security Notes

• API keys stored locally on-device
• No telemetry or analytics — zero servers beyond your AI provider
• Open source audit available via GitHub
• For maximum privacy, use a local LLM endpoint (Ollama) as the AI provider
• System audio capture requires OS-level microphone permissions

FAQ

Q: How is this different from just having a browser window open with ChatGPT?

A: ChatGPT in a browser tab is visible during screen share. Pluely’s compositor-level overlay is invisible to screen capture pipelines. You also get always-on-top positioning, keyboard shortcuts for instant access, and screenshot/audio capture features integrated.

Q: Does it work on Linux?

A: Yes. Pluely ships .deb, .rpm, and .AppImage builds alongside macOS and Windows binaries.

Q: What AI providers does it support?

A: OpenAI (GPT-4o), Anthropic (Claude), Google (Gemini), xAI (Grok), and any custom OpenAI-API-compatible endpoint. Local models via Ollama or LM Studio are supported through the custom provider option.

Q: Is my data sent to Pluely’s servers?

A: No. The app makes direct API calls to your chosen provider. Pluely’s own infrastructure is not involved in AI processing.

Q: How small is it really?

A: The compiled binary is approximately 10MB. By comparison, a typical Electron app is 80-200MB. The Tauri + Rust stack is the key reason for the small footprint.

Conclusion

Pluely fills a specific niche: users who want an AI assistant that stays invisible during meetings without sending data to yet another cloud service. The open-source, GPL-licensed model, small binary size, and direct-API architecture make it a credible privacy-focused alternative to Cluely. For developers or power users already running local models via Ollama, it is a natural pairing.

If you want an always-on-top AI overlay that does not leak into your video calls, Pluely is worth a look. Download it at pluely.com/downloads or browse the source at github.com/iamsrikanthnani/pluely.