Company Policy RAG Chatbot with Pinecone & OpenAI

This n8n workflow implements a Retrieval-Augmented Generation (RAG) chatbot specialized for answering company policy questions. It consists of two core sections: data ingestion and query retrieval. Documents like PDFs from a monitored Google Drive folder are automatically processed, split into chunks, embedded using OpenAI models, and upserted into a Pinecone vector database index.

The ingestion pipeline triggers on new or updated files in Google Drive, downloads them, uses a Recursive Character Text Splitter for optimal chunking with overlap, extracts text via Default Data Loader, generates embeddings, and inserts them into Pinecone. This ensures scalable, searchable storage of policy knowledge.

For user queries, an AI Agent retrieves relevant chunks from Pinecone based on semantic similarity, augments the prompt with this context, and generates precise responses using OpenAI. This setup delivers accurate, hallucination-free answers grounded in your actual policies.

Benefits include effortless knowledge base updates, reduced LLM errors via RAG, and seamless integration of enterprise tools. Ideal for HR teams, compliance officers, or any org needing an internal policy assistant—saving hours on manual searches and ensuring consistent info access.