RAG PDF Q&A: Query Documents with Weaviate & OpenAI

This n8n workflow enables Retrieval-Augmented Generation (RAG) over PDF documents using Weaviate as a vector store and OpenAI for embeddings and chat. It processes a PDF (e.g., a 100+ page arXiv paper), splits it into chunks, generates embeddings, and stores them in Weaviate. Users can then query the content via a chat interface, retrieving relevant chunks and generating precise answers grounded in the document.

The workflow is divided into key parts: manual PDF upload, embedding and indexing into Weaviate, and interactive Q&A using the Question and Answer Chain node. It serves as an extensible template for document-based AI assistants, ideal for analyzing reports, research papers, manuals, or legal docs without manual searching.

Benefits include time-saving automation of document intelligence, reduced hallucinations via RAG, and scalability for custom pipelines. Use cases span research (querying academic papers), business (FAQ from policies), education (textbook Q&A), and content creation (summarizing articles). Requires a Weaviate cluster and OpenAI keys; supports local or cloud setups for self-hosted n8n.

Extend it by automating PDF ingestion from folders/emails or integrating other LLMs/vector DBs for production-grade apps.