README.md

metadata

title: digichat
app_file: app.py
sdk: gradio
sdk_version: 5.27.0
hf_oauth: true

DigiChat

DigiChat is an advanced, modular chatbot framework designed for flexible, multi-agent conversational AI. It supports streaming responses, file processing, integration with Azure AI Search and vector indexes, and orchestration of specialized agents for complex tasks.

Features

• Multi-Agent Architecture:
  Supports multiple agent types, including:

  • BaseAgent: Standard LLM-based conversation.
  • DataAgent: Handles queries involving external data sources.
  • SearchAgent: Integrates with Azure AI Search and vector indexes for semantic search.
  • OrchestratorAgent: Routes and coordinates between specialized agents as tools.

• Streaming Responses:
  Real-time, token-by-token streaming of model outputs for responsive user experience.

• File Upload and Processing:
  Users can upload files (text, images, etc.), which are processed and incorporated into the conversation context.

• Flexible Model Configuration:
  Easily switch between different LLM engines and adjust parameters like temperature.

• Azure AI Search Integration:
  Seamlessly query Azure Search and vector indexes for retrieval-augmented generation.

• Extensible Tooling:
  Add new tools and data sources via configuration, enabling custom workflows.

• History Management:
  Maintains conversation history and supports advanced prompt engineering.

• Robust Error Handling:
  Gracefully handles errors in streaming, file processing, and agent orchestration.

• Logging and Debugging:
  Built-in logging for monitoring and debugging agent behavior.

Usage

1. Start the Chatbot:
   Launch the Gradio interface or integrate DigiChat into your own application.

2. Interact:

   • Send messages and receive streaming responses.
   • Upload files for context-aware answers.
   • Leverage advanced search and data retrieval via specialized agents.

3. Configure:

   • Adjust agent configurations, add tools, or connect new data sources as needed.

Architecture

See architecture.puml for a detailed PlantUML diagram of the system's class structure and relationships.

Requirements

• Python 3.10+
• Gradio
• Azure SDKs (for Search integration)
• Other dependencies as listed in requirements.txt

Setup

1. Clone the repository
2. Install dependencies:

   pip install -r requirements.txt
   

3. Set up your environment variables in .env:

   OPENAI_API_KEY=your_api_key_here
   XAI_API_KEY=your_xai_key_here
   

4. Run the application:

   python app.py
   

License

MIT License

---

For more details, see the code and comments in each module.