A privacy-first, local AI assistant platform for WhatsApp, Telegram, and Ollama. Configure behavior, train the model with knowledge, manage personalities, and control memory — all without touching code.
- Local Ollama-powered AI inference
- WhatsApp companion device bridge integration
- Telegram chat gateway support
- AI identity and personality management
- Global and platform-specific system prompts
- Preset reply engine with exact, keyword, and regex match
- Knowledge base and training import system
- Memory management for short-term and long-term context
- Admin API for runtime configuration
- Backup and restore for AI state
- Secure admin access and health monitoring
Replace these placeholders with real screenshots after setup.
flowchart TD
subgraph Backend
A[FastAPI API] --> B[AI Manager]
A --> C[WhatsApp Gateway]
A --> D[Telegram Gateway]
A --> E[Memory Store]
B --> F[Storage Files]
B --> G[Knowledge Base]
B --> H[System Prompts]
end
subgraph Local AI
I[Ollama] --> A
end
subgraph External
C --> J[WhatsApp Bridge]
D --> K[Telegram API]
end
See INSTALLATION.md and SETUP.md.
- Clone the repository:
git clone https://github.com/naborajs/Self-Hosted-AI-Assistant.git
cd Self-Hosted-AI-Assistant- Install Python dependencies:
python -m pip install -r requirements.txt- Configure your environment in
.env. - Start the WhatsApp bridge:
cd whatsapp_bridge
npm install
npm start- Start the backend:
python main.py- Open
http://127.0.0.1:8000/docsfor API access.
Follow the detailed guide in WHATSAPP_SETUP.md.
Follow the detailed guide in TELEGRAM_SETUP.md.
Follow the detailed guide in OLLAMA_SETUP.md.
See CUSTOMIZATION_GUIDE.md and PERSONALITIES.md.
See KNOWLEDGE_BASE.md.
See ADMIN_API.md.
See DOCKER_GUIDE.md.
See TROUBLESHOOTING.md.
This project is open source under the MIT License.