GraphRAG Panel of Experts AI Chatbot (n8n Workflow Automation)

Give your n8n chatbot smarter answers with GraphRAG “Panel of Experts”

This n8n workflow builds an AI chatbot agent that can query multiple InfraNodus GraphRAG knowledge graphs at once—using a “Panel of Experts” approach to deliver higher-quality responses without setting up complex RAG vector store pipelines.

What this workflow does

• Lets users ask questions through an n8n chat interface (via URL or embeddable UI).
• Automatically selects the best expert knowledge graph: the n8n AI Agent node reviews the list of available tools, where each tool represents a knowledge graph and includes an auto-generated description of what it covers.
• Optionally reformulates the user’s query so it’s better suited for the chosen knowledge graph.
• Queries InfraNodus via HTTP: once the agent decides which expert to use, the workflow sends the query to the InfraNodus HTTP node endpoint to retrieve an answer.

Why GraphRAG instead of standard vector stores?

Unlike typical RAG systems that rely on vector stores, this template uses knowledge graphs (via InfraNodus GraphRAG). That provides:

• Easy, quick setup (no complex data import workflows)
• Holistic knowledge context across your dataset
• Better retrieval of relationships between document chunks for higher-quality responses

Use cases

• Support teams answering customer questions using multiple internal knowledge graphs (policies, troubleshooting, product docs).
• SaaS operators consolidating fragmented documentation into expert-focused knowledge sources.
• Automation engineers building an n8n AI chatbot that routes queries to the most relevant knowledge graph.

Technical details

• n8n AI Agent orchestration (n8n-nodes-langchainagent)
• Chat trigger (n8nnodes-langchainchat trigger)
• LLM via OpenAI chat model (n8nnodes-langchainlm chat open ai)
• Knowledge graph querying through InfraNodus HTTP request tool (http request tool)
• Conversation memory using a window buffer (n8nnodes-langchainmemory buffer window)