Automated Document Ingestion and Q&A with RAG AI

This n8n workflow automates the process of ingesting PDF documents into a vector database and leveraging retrieval-augmented generation (RAG) for answering questions. It monitors a specified Google Drive folder for new PDFs, downloads and processes these files by extracting text, splitting it into manageable chunks, and embedding the data using Cohere embeddings. The embeddings are stored in a Milvus vector database for efficient retrieval. When a chat message is received via webhook, the workflow retrieves relevant information from Milvus, supplies it to an OpenAI GPT-4 model, and responds accordingly. This setup is ideal for organizations needing an intelligent FAQ bot, knowledge base, or document-based AI assistant capable of handling large datasets and multilingual content.