An open-source AI-powered Android home screen launcher with voice control, smart app access, customizable wallpapers, and optional local LLM support.
AI Launcher is an LLM-powered Android home screen launcher that lets users control their phone and launch applications through natural language and voice commands.
- π― Voice First - Control your phone with natural language, hands-free
- π Quick Launch - Intelligently learns your habits, one-tap access to frequently used apps
- π¨ Personalization - Rich themes, wallpapers, and layout customization options
- π Privacy Protection - All data stored locally, never uploaded to the cloud
- Supports Chinese and English voice recognition
- Natural language understanding, control your phone like chatting
- Launch apps, send messages, set alarms with voice
- Offline voice recognition (requires local model download)
- Automatically learns frequently used apps, displayed at the top
- Shows up to 8 frequently used apps (2 rows Γ 4 columns)
- Quick search and launch any app
- App usage statistics and sorting
- Tech Background - Default sci-fi style grid background (can be disabled)
- Image Wallpaper - Choose images from gallery as wallpaper
- Gradient Background - 16 preset gradients + custom colors
- Solid Color - Clean single-color background
- Dark Mode - Auto-follow system or manual toggle
- Supports multiple AI model backends:
- Cloud models (OpenAI, Qwen, DeepSeek, etc.)
- Local models (MLC LLM, completely offline)
- Custom API endpoints and keys
- Detailed permission management
- Rich developer options
- Kotlin - Primary development language
- Jetpack Compose - Modern UI framework
- Material 3 - Latest Material Design specification
- Coroutines - Asynchronous programming
- ViewModel - Lifecycle-aware
- MVVM Pattern - Clear separation of concerns
- Clean Architecture - Layered architecture design
- Repository Pattern - Data access abstraction
- Retrofit + OkHttp - Network requests
- Gson - JSON parsing
- Coil - Image loading
- MLC LLM - Local large language model inference
Go to Releases Page to download the latest APK
# Clone the repository
git clone https://github.com/IceAmber/ai-launcher.git
cd ai-launcher
# Build debug version
./gradlew assembleDebug
# Build release version
./gradlew assembleRelease
# APK located at app/build/outputs/apk/Coming soon to Google Play and other app stores, stay tuned!
- Grant necessary permissions (app list, voice recognition, etc.)
- Choose whether to use tech background (enabled by default)
- Configure AI model (optional)
- Open Settings β Model Configuration
- Select "Cloud Model"
- Enter API endpoint and key
- Supports OpenAI, Qwen, DeepSeek compatible APIs
- Example:
https://api.openai.com/v1
- Open Settings β Model Configuration
- Select "Local Model (MLC)"
- Download model files (required for first use)
- Wait for download to complete
- Tap the microphone icon or say wake word
- Speak commands, for example:
- "Open WeChat"
- "Call Mom"
- "Set alarm for 7 AM tomorrow"
- "What's the weather today"
// Supported API formats
Base URL: https://api.openai.com/v1
API Key: sk-xxx
Model: gpt-3.5-turbo / gpt-4
// Qwen
Base URL: https://dashscope.aliyuncs.com/api/v1
API Key: sk-xxx
Model: qwen-turbo / qwen-plus
// DeepSeek
Base URL: https://api.deepseek.com/v1
API Key: sk-xxx
Model: deepseek-chat- Default model: gemma-2-2b-it
- Model size: ~1.5GB
- Requires Android 5.0+ and 4GB+ RAM
- Completely offline, no network required
- Deep blue background (#0A0E27)
- Grid lines (40dp spacing)
- 7 glowing nodes + connection lines
- Customizable colors and transparency
- Select from gallery
- Adjustable blur level
- Add overlay for better readability
- Custom overlay color and transparency
- 16 preset gradients (Sunset, Ocean, Forest, etc.)
- Custom start and end colors
- Real-time preview
| Permission | Purpose | Required |
|---|---|---|
QUERY_ALL_PACKAGES |
Get app list for display and launch | β Required |
RECORD_AUDIO |
Voice recognition | β Optional |
INTERNET |
Access cloud AI model API | β Optional (not needed for local model) |
VIBRATE |
Haptic feedback | β Optional |
Contributions are welcome! Please feel free to submit issues or pull requests.
We value your privacy:
- β Local Storage - All data saved on device
- β No Data Collection - No personal information collected
- β No Cloud Upload - Voice recognition done locally (when using local model)
- β Open Source Transparency - Code fully open source, auditable
For detailed privacy policy, see PRIVACY.md
Licensed under the Apache License 2.0.
Thanks to these open source projects:
- Jetpack Compose - Modern UI toolkit
- MLC LLM - Local large language model inference
- Coil - Kotlin image loading library
- Retrofit - Type-safe HTTP client
- π§ Email: support@ailauncher.app
- π Website: https://ailauncher.app
- π Issues: GitHub Issues
- π¬ Discussions: GitHub Discussions
If this project helps you, please give it a β Star!
Made with β€οΈ by AI Launcher Team