A privacy-focused, locally-hosted AI companion system with multiple personalities that scales from Raspberry Pi to high-end gaming PCs. ALICE offers different interaction modes while keeping all your conversations and data completely local.
- Assistant Mode: Professional, helpful responses for productivity
- Companion Mode: Casual, friendly conversations with empathy
- Roleplay Mode: Creative character interactions and scenarios
- Dungeon Master Mode: Interactive storytelling with game mechanics
- Storyteller Mode: Narrative-focused creative writing
- Short-term memory: Maintains conversation context
- Long-term memory: Learns and remembers facts about you
- Cross-session continuity: Remembers you between conversations
- Export/import: Take your ALICE with you when upgrading hardware
- Completely local: No data sent to external servers
- Your hardware: Runs on everything from Pi 4B to gaming PCs
- Open source: Full transparency, modify as needed
- No tracking: Conversations stay on your device
- Gaming mode: Temporarily free GPU/RAM for games
- Hardware detection: Automatically selects appropriate AI models
- Scalable performance: Same system works on Pi 4B or RTX 4090
- Web interface: Modern, responsive UI accessible from any device
- Mobile friendly: Bookmark and use like an app on phones/tablets
- Multi-user: Family members can each have conversations
- Network access: Use ALICE from anywhere on your local network
git clone https://github.com/BarbarossaKad/Project-ALICE.git
cd Project-ALICE
python alice_setup.py# Console interface
python alice_core.py
### Console Interface
```bash
python start_console.py
## 📋 Requirements
### Minimum (Raspberry Pi 4B)
- 4GB RAM
- Python 3.8+
- 16GB storage
- Internet for initial model download
### Recommended (Desktop/Gaming PC)
- 8GB+ RAM
- Modern GPU with 6GB+ VRAM (optional but recommended)
- 50GB+ storage for larger models
- Multi-core CPU
### Supported Platforms
- 🐧 **Linux** (Ubuntu, Debian, Pi OS, etc.)
- 🪟 **Windows** (10/11)
- 🍎 **macOS** (Intel/Apple Silicon)
- 🥧 **Raspberry Pi** (4B with 4GB+ RAM)
## 🎯 Supported AI Models
| Hardware | Recommended Models | Response Time |
|----------|-------------------|---------------|
| Pi 4B (4GB) | GPT-2, DistilGPT-2 | 15-30s |
| Pi 4B (8GB) | DialoGPT-Small | 10-20s |
| Desktop (8GB) | DialoGPT-Medium | 5-15s |
| Gaming PC (16GB+) | Mistral-7B, Llama-2-7B | 1-5s |
ALICE automatically detects your hardware and recommends optimal models.
## 🛠️ Installation
### Automatic Setup (Recommended)
```bash
git clone https://github.com/BarbarossaKad/Project-ALICE.git
cd Project-ALICE
python alice_setup.pyThe setup script will:
- Detect your hardware capabilities
- Install appropriate dependencies
- Download recommended AI model
- Create configuration files
- Set up directory structure
# Clone repository
git clone https://github.com/BarbarossaKad/Project-ALICE.git
cd Project-ALICE
# Create virtual environment
python -m venv alice_env
source alice_env/bin/activate # Linux/Mac
# alice_env\Scripts\activate # Windows
# Install dependencies
pip install -r requirements.txt
# Run setup
python alice_setup.pydocker-compose up -d
# Access at http://localhost:7860# Switch personality modes
/mode companion
/mode storyteller
/mode dungeon_master
# System management
/gaming # Free resources for gaming
/resume # Resume after gaming
/memory # View conversation history
/status # System information
/help # Full command list- Start web interface:
python alice_web_interface.py - Open browser to
http://localhost:7860 - Chat with ALICE using the modern UI
- Switch modes, view memory, export conversations
- Find your computer's IP address
- Bookmark
http://192.168.1.xxx:7860on your phone - Add to home screen for app-like experience
Edit alice_config.yaml:
current_mode: companion
model_name: "microsoft/DialoGPT-medium"
temperature: 0.7
max_tokens: 200
safety_level: moderateCreate custom personas in personas/custom_personas.json:
{
"teacher": {
"name": "ALICE Teacher",
"personality": "Educational, patient, encouraging",
"style": "Clear explanations with examples",
"restrictions": ["Keep content age-appropriate", "Focus on learning"]
}
}The model manager automatically optimizes settings based on your hardware, but you can override:
# For high-end GPUs
config = {
"model_name": "mistralai/Mistral-7B-Instruct-v0.1",
"load_in_8bit": True,
"device_map": "auto"
}┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Web Interface │ │ Core System │ │ Model Manager │
│ (Gradio UI) │◄──►│ (alice_core) │◄──►│ (AI Models) │
└─────────────────┘ └──────────────────┘ └─────────────────┘
│
┌──────────────────┐
│ Memory System │
│ (SQLite DB) │
└──────────────────┘
- Core System: Main ALICE logic, personality management
- Memory System: Conversation storage and user learning
- Model Manager: AI model loading, optimization, scaling
- Web Interface: Modern UI accessible from any device
- Persona System: Switchable AI personalities and behaviors
- Installation Guide - Detailed setup instructions
- API Reference - Technical documentation
- Contributing - How to contribute to the project
- FAQ - Common questions and troubleshooting
Contributions are welcome! Whether you're:
- 🐛 Reporting bugs
- 💡 Suggesting features
- 📖 Improving documentation
- 🔧 Adding new personalities
- ⚡ Optimizing performance
Please read CONTRIBUTING.md for guidelines.
- Languages: Python, JavaScript, HTML/CSS
- AI Framework: Hugging Face Transformers, PyTorch
- UI Framework: Gradio
- Database: SQLite
- Supported Models: 15+ pre-configured AI models
- Platforms: Cross-platform (Linux, Windows, macOS, Pi OS)
- Voice input/output support
- Plugin system for custom extensions
- Multi-language support
- Advanced memory search and organization
- Integration with external APIs (optional)
- Mobile apps (iOS/Android)
- Advanced roleplay scenarios
- Educational mode for learning
- Home automation integration
- Multi-user conversation rooms
This project was developed with assistance from AI collaborators:
- Claude (Anthropic): Primary development assistance, system architecture, and technical implementation
- ChatGPT (OpenAI): Initial concept development, technical specifications, and documentation support
This project is licensed under the MIT License - see the LICENSE file for details.
- ✅ Commercial use - Use ALICE in commercial projects
- ✅ Modification - Adapt and change the code as needed
- ✅ Distribution - Share with others
- ✅ Private use - Use for personal projects
⚠️ Attribution required - Credit the original project
- Hugging Face - For the transformers library and model ecosystem
- Gradio - For making beautiful web interfaces simple
- PyTorch - For the underlying AI framework
- The open-source community - For inspiring privacy-focused AI development
- Teachers and educators - Who inspired the educational focus and accessibility
- Issues: GitHub Issues
- Discussions: GitHub Discussions
- Wiki: Project Wiki
If you find ALICE helpful, please consider giving it a star! ⭐