RAG Chatbot for Company Docs with Google Drive & Gemini

This n8n workflow creates a Retrieval Augmented Generation (RAG) chatbot tailored for company knowledge bases. It automatically detects new or updated documents in a Google Drive folder using two trigger nodes, downloads them, extracts content, splits into chunks, generates embeddings with Google's Gemini text-embedding-004 model, and stores them in a Pinecone vector database. This ensures the knowledge base stays current without manual intervention.

The chat interface is powered by a Chat Trigger node feeding into an AI Agent. The agent employs a Vector Store Tool connected to Pinecone in query mode to retrieve relevant document chunks based on user questions. These chunks, combined with the query, are processed by the Gemini Pro chat model to generate precise, informative answers grounded in company documents.

Key benefits include reduced hallucination in AI responses, real-time indexing of updates, conversational memory via Window Buffer Memory for context-aware chats, and seamless integration of enterprise tools. Ideal use cases: HR query bots for policies, sales teams accessing product docs, or engineering support from technical manuals, saving hours on information retrieval and boosting productivity.