Website RAG Chatbot with Pinecone, Airtable & OpenAI

This n8n workflow automates the creation of a powerful RAG (Retrieval-Augmented Generation) chatbot for website content. It starts by fetching content from a specified website via HTTP Request, extracts the HTML body, converts it to clean Markdown, and splits the text into manageable chunks (around 500 characters with overlap) using a Character Text Splitter. OpenAI generates embeddings for each chunk, which are then stored in Pinecone with namespace support for organized indexing.

The chatbot component leverages a Chat Agent powered by OpenAI or OpenRouter, retrieving relevant chunks from Pinecone based on user queries. It includes a memory buffer for multi-turn conversations, ensuring context-aware responses. Airtable integration dynamically pulls in billing or other structured data, enhancing answers for e-commerce scenarios like order status or pricing.

Benefits include rapid knowledge base creation from websites, cost-effective customer support automation, and scalable RAG for FAQs, product info, or docs. Ideal for e-commerce order processing, it saves hours of manual data handling and boosts query accuracy over generic LLMs. Use cases: customer Q&A bots, support agents, or internal knowledge retrieval.