LLM Wherever - Large Language Model Client for Apple Watch

A LLM client application designed specifically for Apple Watch, allowing you to chat with AI assistants anytime, anywhere on your watch.

Features

📱 iPhone App

• Smart API Configuration: Add and manage multiple API providers (OpenAI, Claude, SiliconFlow, etc.)
• Automatic Model Fetching: Automatically fetch available model lists after entering API Key
• Preset Templates: Built-in configuration templates for popular APIs like OpenAI, Claude, SiliconFlow
• Secure Storage: API Keys are securely stored on local device
• Real-time Sync: Configuration automatically syncs to Apple Watch

⌚ Apple Watch App

• Ultimate Optimization: Compact interface designed specifically for small screens
• Liquid Glass Style: Featuring Apple's latest translucent material design
• Smart Conversations: Support for multi-turn conversations with context preservation
• Markdown Support: Rich text rendering for AI responses with syntax highlighting
• Haptic Feedback: Rich haptic feedback enhances interaction experience
• Voice Input: Support for voice-to-text input (system-level support)
• Smooth Animations: Smooth scrolling and transition animations
• Quick Response: Optimized network requests and error handling
• Local Storage: Chat history securely stored locally on the watch

Screenshots

📱 iPhone App

Home	Provider Details	Settings

⌚ Apple Watch App

Chat History	Chat Interface	Settings

Supported API Providers

• OpenAI: GPT-4, GPT-4o, GPT-3.5 Turbo
• Claude: Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Haiku
• SiliconFlow: Qwen2.5-7B-Instruct, Qwen2.5-72B-Instruct, DeepSeek-V2.5
• Custom APIs: Support for other APIs compatible with OpenAI format

Technical Architecture

iOS App

• SwiftUI: Modern user interface framework
• WatchConnectivity: Data synchronization between iPhone and Apple Watch
• UserDefaults: Local data persistence
• Async/Await: Modern asynchronous processing

watchOS App

• Native SwiftUI: Interface optimized specifically for watch
• MarkdownUI: Rich text rendering using swift-markdown-ui
• Memory Optimization: Optimized for watch hardware limitations
• Low Power Design: Minimized battery consumption
• Adaptive Layout: Support for different sizes of Apple Watch

How to Use

1. Configure API (iPhone)

1. Open the "LLM Wherever" app on iPhone
2. Tap "Add API Provider"
3. Select a preset provider (OpenAI, Claude, SiliconFlow) or create custom configuration
4. Enter API Key
5. Tap "Test Connection and Fetch Models" to automatically get available models
6. Configuration will automatically sync to Apple Watch

2. Start Conversation (Apple Watch)

1. Open "LLM Wherever" app on Apple Watch
2. Select API provider and model for first use
3. Input text or use voice input in the optimized chat interface
4. Experience smooth conversation interaction and haptic feedback
5. AI responses support Markdown formatting (bold, italic, code blocks, lists, etc.)
6. Long press send button to stop ongoing requests
7. Tap the gear icon in top right corner to change model anytime

Design Philosophy

Authentic Apple Ecosystem Experience

• Strictly follows Apple Human Interface Guidelines
• Uses native system UI components
• Maintains consistent interaction patterns with the system
• Supports Dark Mode and accessibility features

Liquid Glass Design Style

• Translucent effects and material textures
• Smooth animations and transition effects
• Modern rounded corners and shadows
• Clear visual hierarchy with distinct layers

User Experience First

• Simple and intuitive operation flow
• Quick response to interaction feedback
• Smart error handling and prompts
• Seamless synchronization experience between devices

Privacy and Security

• Local Storage: All API Keys and chat records are stored only on user devices
• End-to-End Encryption: Data transmission between iPhone and Apple Watch is encrypted
• No Cloud Service: The app itself does not collect or store any user data
• Transparent Communication: Only communicates with API endpoints configured by users

System Requirements

• iPhone: iOS 17.0+
• Apple Watch: watchOS 10.0+
• Pairing Requirement: Requires iPhone and Apple Watch to be paired
• Network Connection: Requires internet connection to access LLM APIs

Dependencies

This project uses the following third-party libraries:

• swift-markdown-ui: For rendering Markdown text in SwiftUI on Apple Watch

Adding Dependencies

To build this project, you need to add swift-markdown-ui to your Xcode project:

1. In Xcode, go to File → Add Package Dependencies
2. Enter the URL: https://github.com/gonzalezreal/swift-markdown-ui
3. Select version 2.0.2 or later
4. Add MarkdownUI to your LLM Wherever Watch App target

Development Notes

Project Structure

LLM Wherever/
├── LLM Wherever/                 # iPhone App
│   ├── Models/                   # Data Models
│   ├── Managers/                 # Manager Classes
│   ├── Services/                 # Service Layer
│   ├── Views/                    # View Components
│   └── ContentView.swift         # Main View
└── LLM Wherever Watch App/       # Apple Watch App
    ├── Models/                   # Data Models (Shared)
    ├── Managers/                 # Manager Classes
    ├── Services/                 # Service Layer
    ├── Views/                    # View Components
    └── ContentView.swift         # Main View

Key Components

• APIProvider: API provider data model
• WatchConnectivityManager: Inter-device communication management
• LLMService: API calling service
• ChatMessage: Message data model

Reference Projects

This project references design concepts from the following open source projects:

• humane.watch - AI assistant on Apple Watch
• WatchGPT2 - Locally running GPT2
• PlaticaBot - Cross-platform ChatGPT client

License

MIT License - See LICENSE file for details

Contributing

Issues and Pull Requests are welcome to improve this project!