Chat with Google Drive documents using Pinecone and OpenAI RAG

Google Drive → Pinecone RAG Chatbot (Auto-Sync & Query)

This n8n workflow implements a fully automated Retrieval-Augmented Generation (RAG) pipeline powered by Google Drive, OpenAI embeddings, and Pinecone.

It continuously keeps a vector database in sync with your company documents and exposes them through an AI chat interface.

What this workflow does

The workflow monitors a Google Drive folder and automatically reacts to document lifecycle events:

File created

File updated

File deleted

When a document is added or updated:

The file is downloaded from Google Drive

Its content is chunked using a recursive text splitter

Embeddings are generated with OpenAI

Vectors are stored or updated in Pinecone

When a document is deleted:

The corresponding vectors are removed from Pinecone, keeping the index clean and consistent

On the chat side:

A conversational AI agent retrieves relevant vectors from Pinecone

Context is injected into the prompt

The assistant answers questions grounded only on your documents

Key features

End-to-end RAG pipeline (ingestion + retrieval + chat)

Automatic vector updates on file changes

Idempotent design (safe re-runs, no duplicated vectors)

Google Drive as a live knowledge source

Pinecone as scalable vector storage

OpenAI embeddings and chat models

Ready-to-use AI chat interface inside n8n

Typical use cases

Internal company knowledge base

AI assistant for policies, manuals, and documentation

Team chat over shared Google Drive files

Lightweight alternative to full-blown document search platforms

Prototyping and production RAG systems

Who this template is for

n8n users building AI-powered workflows

Teams working with Google Drive documents

Developers implementing RAG architectures

Anyone who wants a self-hosted, controllable, and transparent AI document chatbot

This template is designed to be robust, maintainable, and production-ready, while remaining easy to extend with additional data sources, metadata filtering, or alternative LLM providers.