Your Personal AI Assistant - Private, Powerful, and Always Ready

MyDeviceAI brings the power of artificial intelligence directly to your mobile device, with a focus on privacy, performance, and personalization. Run AI models locally on your iOS or Android device, or connect remotely to your desktop computer from anywhere using WebRTC.

What makes MyDeviceAI unique: We're the only local AI app that lets you seamlessly run powerful desktop models remotely on your computer via secure peer-to-peer WebRTC connections. Use your mobile device as a gateway to desktop-class AI models through MyDeviceAI-Desktop, all while maintaining complete privacy and control.

By combining on-device AI processing with optional remote desktop connectivity and privacy-focused web services, MyDeviceAI offers the best of all worlds: the convenience of mobile, the power of desktop, and the privacy of local-first computing.

• Seamless Experience: The app comes with bundled AI models that load asynchronously in the background, ensuring zero waiting time and a smooth user experience from the moment you launch.

• Remote Desktop Connection: Connect your mobile device to MyDeviceAI-Desktop using secure peer-to-peer WebRTC connections. Leverage the power of desktop-class AI models directly from your iOS or Android device with automatic retry and reconnection support.

• Web Search Integration: Bridge the knowledge gap between local AI and cloud capabilities with our web search integration, utilizing a self-hosted SearXNG instance for privacy-focused results.

• Thinking Mode: Powered by Qwen 3, our Thinking Mode helps you tackle complex problems and brainstorming sessions with advanced reasoning capabilities.

• Personalization (Beta): Create a truly personalized AI experience by loading saved user contexts. Tailor conversations to your preferences and needs (Currently in Beta).

• Chat History: Access up to 30 days of conversation history, making it easy to reference past discussions and maintain context over time.

• Broad Device Support: Compatible with all modern iPhones and Android devices, bringing private AI capabilities to more users than ever before.

• macOS
• Xcode 13.0+
• CocoaPods
• Node.js 14.0 or later
• React Native development environment set up (React Native - Environment Setup)

1. Clone the repository:

   git clone https://github.com/navedmerchant/MyDeviceAI.git
   cd MyDeviceAI

2. Download required AI models:

   • Download Qwen3 1.7B Q4 and place it in the ios/ directory
   • Download BGE Small and place it in the ios/ directory

3. Create configuration file:

   • Create src/config/Env.ts with the following content:

     export const HOSTNAME = 'YOUR_SEARXNG_INSTANCE_URL';
     export const P2PCF_WORKER_URL = 'YOUR_P2PCF_SIGNALLING_SERVER_URL';

   Important Configuration Notes:

   • HOSTNAME: URL of your SearXNG instance for web search functionality (optional, but recommended for web search features)
   • P2PCF_WORKER_URL: URL of your P2PCF signalling server (required for remote desktop connections)
     • This enables secure peer-to-peer WebRTC communication between mobile and desktop clients
     • Without this, remote desktop features will not work
     • Example: https://your-p2pcf-server.com/
     • Deployment Options:
       • Option 1 (Cloudflare Workers): Deploy p2pcf worker.js on Cloudflare Workers (free tier available)
       • Option 2 (Railway): Deploy p2pcf-signalling on Railway (easy one-click deployment)

4. Install dependencies:

   npm install
   # or
   yarn install

5. Install pods:

   cd ios && pod install && cd ..

6. Start the application:

   npm run ios
   # or
   yarn ios

MyDeviceAI supports secure peer-to-peer connections between mobile devices (iOS/Android) and desktop computers, allowing you to access powerful desktop-class AI models from your phone.

The remote connection feature uses WebRTC technology through the p2pcf.rn library to establish direct peer-to-peer connections:

• Cross-Platform Support: Works seamlessly on both iOS and Android devices
• Secure Communication: End-to-end encrypted WebRTC data channels
• 6-Digit Room Codes: Simple room-based pairing system
• Automatic Retry: Built-in connection retry with exponential backoff (up to 3 attempts)
• Real-time Streaming: Supports streaming responses with both regular and reasoning tokens

1. P2PCF Signalling Server: You need a P2PCF signalling server for WebRTC coordination. Configure the URL in src/config/Env.ts:

   export const P2PCF_WORKER_URL = 'https://your-p2pcf-server.com/';

2. MyDeviceAI-Desktop: Install and run the MyDeviceAI-Desktop companion app on your computer

1. Start MyDeviceAI-Desktop on your computer and generate a 6-digit room code
2. Open MyDeviceAI mobile app
3. Navigate to Remote Connection settings
4. Enter the 6-digit room code
5. Once connected, your mobile app will use the desktop's AI models for inference

• Protocol Handshake: Automatic version negotiation ensures compatibility between mobile and desktop clients
• Connection Monitoring: Real-time connection status with retry countdown
• Error Handling: Comprehensive error handling with user-friendly messages
• Automatic Reconnection: Automatically attempts to reconnect on disconnection (up to 3 attempts with exponential backoff)
• OpenAI-Compatible API: Uses standard message format for seamless integration

• Protocol Version: 1.0.0
• Transport: WebRTC Data Channels
• Message Format: JSON-based P2P protocol
• Retry Strategy: Exponential backoff (2s → 3s → 4.5s, max 30s)
• Max Retry Attempts: 3
• Polling Interval: 3 seconds (mobile polls for desktop availability)

• Framework: React Native (iOS/Android)
• AI Models:
  • Qwen 3 (1.7B Q4) for chat and reasoning
  • BGE Small for embeddings
• Remote Connection:
  • p2pcf.rn for WebRTC peer-to-peer communication
  • Custom P2P protocol for desktop-mobile sync
• Web Search: SearXNG Integration
• State Management: Redux + Redux Toolkit
• Storage: AsyncStorage for chat history and user contexts
• UI Components: Lucide
• Code Quality: ESLint, Prettier
• Build Tools: Metro Bundler

We welcome contributions to MyDeviceAI! Here's how you can help:

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

• Follow the React Native community best practices
• Write meaningful commit messages
• Update documentation for any new features
• Add appropriate tests for new functionality
• Ensure privacy-first approach in feature implementation

MyDeviceAI is built with privacy at its core:

• All AI processing happens locally on your device
• No data is sent to external servers without explicit consent
• Web Search integration via SearXNG respects user privacy
• Chat history is stored locally and never uploaded

This project is licensed under the MIT License - see the LICENSE file for details.

If you discover any bugs, please create an issue in the GitHub repository including:

• iOS version
• iPhone model
• Steps to reproduce the bug
• Expected and actual behavior

We are actively working on the following issues and welcome contributions:

1. Formula Rendering: Need to implement mathematical formula support using KaTeX or a Redux-compatible parser
2. Image Rendering: Some base64 images from search results fail to render properly
3. Think Token Display: Occasional display of raw think tokens in the thinking section needs better parsing logic

If you'd like to help solve any of these issues, please check our Contributing section above!

If you have any questions or suggestions, feel free to open an issue or reach out to the maintainers.

Made with ❤️ by Naved Merchant