n8n RAG Chatbot: Ollama + Qdrant Retrieval QA Workflow

Build a local RAG chatbot in n8n that answers from your own Qdrant knowledge base

This n8n workflow lets you create a retrieval-augmented generation (RAG) chat experience on your machine: it pulls relevant passages from a Qdrant vector collection, then generates an answer with a local Ollama model—staying grounded in the retrieved context.

What this workflow does

• Receives your question via an n8n chat trigger.
• Retrieves relevant document chunks from a Qdrant collection using an Ollama-powered embeddings model.
• Injects retrieved context into a retrieval QA prompt that instructs the assistant to answer only from the provided passages and to decline when information is missing.
• Generates a concise response using the local Ollama chat model qwen2.5:7b.

Use cases

• Internal support chatbot for your handbook or SOPs, grounded in your own indexed documents.
• Knowledge Q&A for SaaS operations (policies, troubleshooting notes, release procedures) with transparent, context-only answers.
• RAG testing in automation projects—tune retrieval by increasing Top K when multi-fact questions need broader context.

Technical details

• n8n trigger: n8nn8n-nodes-langchainchat chat trigger
• Embeddings: n8nn8n-nodes-langchainembeddings with Ollama (e.g., nomic-embed-text:latest)
• Vector store: n8nn8n-nodes-langchainvectorstore connected to Qdrant
• Retrieval QA: n8nn8n-nodes-langchainchain retrieval QA chain
• LLM: n8nn8n-nodes-langchainlm Ollama chat model (qwen2.5:7b)

Setup essentials

• Run the companion indexing workflow first: “Index local documents for RAG using Ollama embeddings and Qdrant” (https://creators.n8n.io/workflows/17764).
• Start Ollama locally and pull qwen2.5:7b and nomic-embed-text:latest.
• Create n8n credentials for local Ollama and your Qdrant instance.
• Set the Qdrant collection name to match the one used during indexing (e.g., “handbook”).