Description

A WebSocket-based AI backend service that powers the conversational AI assistant integrated into the Trento web UI. The service assists SAP administrators in monitoring and managing SAP Infrastructure through natural language interactions. It combines direct system access via the Trento MCP (Model Context Protocol) server with a RAG (Retrieval-Augmented Generation) system that provides context from Trento and SUSE documentation. Built on LangGraph's multi-agent architecture, it maintains WebSocket connections from the Trento UI, enabling real-time conversational queries about cluster health, check results, and best practices while preserving conversation context across multiple interactions.

Goals

• Enable natural language interface for SAP Infrastructure monitoring and management
• Process real-time queries from the UI with streaming responses for better user experience
• Deliver accurate, context-aware responses by combining real-time system data with documentation
• Support both technical operations (cluster status checks) and knowledge queries (best practices)
• Demonstrate practical integration of the MCP protocol within existing enterprise applications

Roadmap

• Develop a React-based AI assistant widget integrated into the Trento web UI
• Provide WebSocket backend service for the Trento UI's AI assistant widget
• Deploy all the components on a Kubernetes cluster for testing and demonstration
• Record demo videos showcasing the AI assistant's capabilities in real-world scenarios

Resources

• Trento Project: https://github.com/trento-project/web - SAP infrastructure monitoring
• Rancher AI Agent: https://github.com/rancher-sandbox/rancher-ai-agent - Original multi-agent architecture inspiration
• Model Context Protocol: https://modelcontextprotocol.io - MCP specification and documentation
• Vector database : for RAG implementation, such as Chroma or Milvus - Vector database for RAG implementation
• LangGraph: https://github.com/langchain-ai/langgraph - Multi-agent workflow orchestration framework
• LangChain: https://github.com/langchain-ai/langchainjs - Framework for LLM applications