Document-to-Knowledge-Base Automation (n8n + Pinecone RAG)

I built an n8n automation that turns uploaded business documents into a searchable AI knowledge base. It’s the data pipeline behind a RAG (retrieval-augmented generation) chatbot or support assistant.

Use case: Businesses keep their information in scattered files like FAQs, service guides, price lists, policies and spreadsheets. This workflow lets someone upload those files through a simple form, and it prepares them automatically so an AI assistant can answer customer or team questions from the company’s own content.

Files are submitted through an n8n form (CSV, PDF, DOCX or TXT)

A JavaScript step and a Switch node detect each file type and send it to the right extraction path

The text is extracted, cleaned and structured, with metadata attached

The text is split into chunks, converted to vector embeddings with OpenAI, and stored in a Pinecone vector database

The knowledge base can be refreshed or cleared through the Pinecone API

Result: Multiple files in different formats are processed in a single run and become searchable by meaning, not just keywords. I tested it with a business’s service documentation, which was split into 57 searchable chunks.

Best for: customer support bots, internal company assistants, FAQ automation, and any business that wants an AI that answers from its own documents.

Stack: n8n, Pinecone, OpenAI Embeddings, JavaScript, REST APIs

For the skills tags, use: n8n, Workflow Automation, RAG, Pinecone, Vector Database, OpenAI, Document Processing, AI Chatbot.