Jarvis - An Advanced AI Assistant

Jarvis is an experimental AI assistant that explores various approaches to artificial intelligence, natural language processing, and cognitive architectures. Built as a Telegram bot, it serves as a platform for testing different LLM models, frameworks, and cognitive architectures.

Key Features

• 🤖 Advanced conversation capabilities using Azure OpenAI
• 🏠 Home Assistant integration for IoT control
• 📈 Stock analysis and visualization
• 🔍 Web search and information retrieval
• 🧠 Memory and context persistence
• 🤔 Internal thought processes and reasoning

Project Branches

The project contains several experimental branches, each exploring different aspects of AI:

• main: Stable release with core functionality
• home-assistant: Integration with Home Assistant for IoT control
• autogen: Implementation using Microsoft's AutoGen framework
• autogen_memory: Enhanced memory capabilities using AutoGen
• central_thoughts: Exploration of internal thought processes
• langchain: Integration with LangChain framework
• semantic-kernel: Microsoft Semantic Kernel implementation
• tools: Various tool integrations and utilities
• stock-images: Stock market analysis and visualization features

Prerequisites

• Python 3.10 or higher
• Poetry for dependency management
• Azure OpenAI API access
• Telegram Bot Token (from BotFather)
• (Optional) Home Assistant instance for IoT control

Installation

1. Clone the repository:

   git clone https://github.com/yusufk/jarvis-azure.git
   cd jarvis-azure

2. Install dependencies using Poetry:

   poetry install

3. Create a .env file in the project root with required environment variables:

   TELEGRAM_TOKEN=your_telegram_bot_token
   AZURE_OPENAI_API_KEY=your_azure_openai_key
   AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint
   # Optional for Home Assistant integration
   MCP_HOST_URL=your_home_assistant_url
   MCP_AUTH_TOKEN=your_home_assistant_token

Configuration

1. Set up your Azure OpenAI deployments:

   • Create a deployment for chat completion (e.g., GPT-4)
   • Configure the deployment name in your environment variables

2. For Home Assistant integration (optional):

   • Ensure your Home Assistant instance is accessible
   • Generate a long-lived access token from Home Assistant
   • Configure the URL and token in your environment variables

Running the Bot

1. Start the bot:

   poetry run python jarvis.py

2. Available commands in Telegram:

   • /start - Initialize the bot
   • /clear - Clear conversation history
   • /status - Check bot status
   • Just chat normally for other interactions

Features in Detail

Core Capabilities

• Natural conversation using Azure OpenAI's models
• Persistent memory across conversations
• Contextual understanding and response generation
• Tool integration for extended functionality

Home Assistant Integration

• Control smart home devices
• Query device states
• Execute automation scripts
• Monitor sensors and systems

Stock Analysis

• Real-time stock price analysis
• Historical data visualization
• Trend analysis and basic predictions

Cognitive Architecture

• Internal thought processes
• Self-reflection capabilities
• Memory management and context awareness
• Tool usage reasoning

Contributing

Contributions are welcome! Feel free to:

• Submit bug reports or feature requests through issues
• Create pull requests for improvements
• Experiment with new branches for different approaches

License

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