A powerful multiplatform desktop application that provides universal AI assistance through global hotkeys. Access any LLM model from anywhere on your system - Word documents, IDEs, browsers, or any text field. Built with Tauri 2.0 (Rust + TypeScript + Tailwind CSS).
- Universal Hotkey: Press
Ctrl+Space(configurable) from any application - Smart Text Capture: Automatically detects selected text with clipboard fallback
- System Tray Integration: Runs minimized with right-click context menu
- Streaming Responses: Real-time token-by-token output with cancel support
- Conversation History: Searchable, persistent history with re-run capability
- Custom Tasks Library: Create your own AI tasks with dynamic options
- Multiple Response Modes: Auto-paste, clipboard mode, or review mode
- Encrypted Configuration: API keys stored securely with AES-GCM encryption
- Markdown & LaTeX: Full rendering support including syntax highlighting and KaTeX math
| Task | Description |
|---|---|
| Custom Task | Flexible AI help for any question |
| Email Reply | Professional replies with tone/length options |
| Text Summarization | Condense text with format options |
| Text Translation | 14+ languages supported |
| Text Rewriting | Improve text with configurable tones |
| Image Generation | DALL-E, FLUX.1, or compatible models |
| Speech-to-Text | Whisper-compatible transcription |
| Text-to-Speech | Natural-sounding audio generation |
| Unicode Symbols | Quick access to special characters |
| WhatsApp Response | Casual message replies |
- Download from Releases (
.msifor Windows,.AppImagefor Linux) - Run the installer - AI Anywhere appears in your system tray
- Right-click tray icon → Settings to configure
git clone https://github.com/bgeneto/AIAnywhere.git
cd AIAnywhere
npm install
npm run dev # Development mode
npm run tauri build # Production buildDebian/Ubuntu
sudo apt install libwebkit2gtk-4.1-0 libayatana-appindicator3-1 libxdo3Fedora/RedHat
sudo dnf install webkit2gtk4.1 libayatana-appindicator3 libxdoWayland: Ensure you have a compatible portal (e.g., xdg-desktop-portal-gnome) for global shortcuts.
- Right-click the system tray icon → Settings
- Configure:
- API Base URL: Your LLM endpoint (e.g.,
https://api.openai.com/v1) - API Key: Your API key (encrypted automatically)
- LLM Model: Click "Get Models" and select one
- API Base URL: Your LLM endpoint (e.g.,
- Save and start using!
- Select text in any application (optional)
- Press hotkey (
Ctrl+Space) - Choose task and configure options
- Click Send or press
Ctrl+Enter
Works with any OpenAI-compatible API. Auto-detects available models.
| Provider | Base URL | Notes |
|---|---|---|
| OpenAI | https://api.openai.com/v1 |
Full support |
| Anthropic | https://api.anthropic.com/v1 |
Text only |
| Google Gemini | https://generativelanguage.googleapis.com/v1beta/openai/ |
Text + vision |
| Ollama | http://localhost:11434/v1 |
Local LLMs |
| LM Studio | http://localhost:1234/v1 |
Local LLMs |
| OpenRouter | https://openrouter.ai/api/v1 |
100+ models |
| Together AI | https://api.together.xyz/v1 |
Open-source models |
| Azure OpenAI | https://<resource>.openai.azure.com/v1 |
Enterprise |
| Setting | Default | Description |
|---|---|---|
hotkey |
Ctrl+Space |
Global activation |
pasteBehavior |
reviewMode |
autoPaste, clipboardMode, or reviewMode |
disableTextSelection |
false |
Skip auto text capture for faster startup |
enableDebugLogging |
false |
Log API requests |
API Key not working
- Verify key is correct (no spaces)
- Check endpoint URL format
- Enable
enableDebugLoggingto see errors - Verify account has credits
Text selection not working
- Copy text manually (
Ctrl+C) before hotkey - Use "Clipboard Mode" in Settings
- Some protected apps block text capture
Paste not working
- Try "Review Mode" first
- Use "Clipboard Mode" for protected fields
- Restart app if paste stops responding
Linux: Application won't start
Install missing dependencies:
# Debian/Ubuntu
sudo apt install libayatana-appindicator3-1 libxdo3 libwebkit2gtk-4.1-0
# Fedora
sudo dnf install libayatana-appindicator3 libxdo webkit2gtk4.1For Wayland + NVIDIA issues:
WEBKIT_DISABLE_DMABUF_RENDERER=1 ai-anywhereLinux Wayland: Global hotkey and text capture limitations
Due to Wayland's security model, the following limitations apply:
-
Global hotkeys may not work: Wayland prevents applications from registering global shortcuts directly. You may need to configure a system-level shortcut in your desktop environment (GNOME/KDE Settings → Keyboard Shortcuts) to launch AI Anywhere.
-
Automatic text capture is not possible: Wayland blocks applications from simulating keystrokes (like Ctrl+C) in other apps. You must manually copy (Ctrl+C) your selected text before triggering the hotkey.
Recommended workflow on Wayland:
- Select text in any application
- Press
Ctrl+Cto copy - Press your system shortcut to open AI Anywhere (or use the tray icon)
- The copied text will automatically appear in the prompt
Note: These are fundamental Wayland security restrictions, not bugs. Consider using an X11 session if you need full global hotkey support.
- Node.js 18+
- Rust 1.70+ (rustup.rs)
- Platform build tools (VS Build Tools / XCode / build-essential)
AIAnywhere/
├── src/ # React frontend
│ ├── components/ # UI components
│ ├── context/ # AppContext state
│ ├── i18n/ # Translations (en, pt-BR)
│ └── types/ # TypeScript definitions
├── src-tauri/ # Rust backend
│ └── src/
│ ├── lib.rs # Tauri commands
│ ├── llm.rs # API communication
│ ├── operations.rs # Task definitions
│ └── config.rs # Configuration
└── package.json
npm run dev # Development with hot reload
npm run build # Production build
npm run tauri build # Create installers
cargo fmt && cargo clippy # Format/lint Rust- Add to
OperationTypeenum inoperations.rs - Define options and system prompt in
get_operations() - Update TypeScript type in
src/types/index.ts
- Fork the repository
- Create feature branch (
git checkout -b feature/Amazing) - Commit changes (
git commit -m 'Add Amazing') - Push and open a Pull Request
See Architecture Overview for detailed guidelines.
Q: Is my API key safe?
A: Yes. Encrypted with AES-GCM-256, stored locally, never transmitted except to your configured endpoint.
Q: Does it work on macOS/Linux?
A: Yes! Build from source, or use Linux releases. macOS binaries coming soon.
Q: Can I customize prompts?
A: Yes. Edit config.json directly with placeholders like {tone}, {language}.
Q: What data is collected?
A: None. Zero telemetry. Data only goes to your configured LLM API.
MIT License - Copyright (c) 2025 LABiA-FUP/UnB
Made with ❤️ for productivity enthusiasts who want AI assistance everywhere they work.