Local RAG Chatbot: PDF Q&A with Ollama & Qdrant

This n8n workflow enables building a fully local AI chatbot powered by Retrieval Augmented Generation (RAG). It uses a form trigger to upload PDF files, processes them by chunking the content, generating embeddings with mxbai-embed-large via Ollama, and storing them in a Qdrant vector database collection named 'rag_collection'. A separate chatbot interface retrieves relevant document chunks based on user queries, augments the prompt, and generates accurate, context-aware responses using Llama 3.2.

Key benefits include complete data privacy with no cloud dependencies, cost-free operation after setup, and scalability for handling multiple PDFs. It's ideal for offline environments, reducing latency and avoiding API costs associated with cloud LLMs or vector stores.

Use cases span enterprise knowledge bases for internal document Q&A, research assistants analyzing academic papers, legal/contract review bots, customer support from product manuals, and personalized learning tools from educational materials. Setup requires the n8n self-hosted AI starter kit, making it accessible for self-hosters seeking powerful, local AI automation.