n8n Telegram RAG Order Bot: Google Drive to Supabase Vector

Turn new Google Drive files into a Supabase-powered RAG knowledge base for your Telegram orders bot

This n8n workflow watches a specific Google Drive folder, converts newly added documents into searchable text chunks, creates OpenAI embeddings, and stores them in a Supabase Vector Store—so your Telegram orders bot can retrieve accurate answers using retrieval-augmented generation (RAG).

What this workflow does

• Triggers every minute when a new file is created in a configured Google Drive folder.
• Downloads the newly added file from Google Drive.
• Extracts plain text from the downloaded file.
• Splits the text for retrieval into 300-character chunks with 20-character overlap to improve match quality.
• Generates embeddings for each chunk using OpenAI (model can be changed).
• Inserts chunks + embeddings into a Supabase Vector Store table (e.g., a table named documents).

Use cases

• Build a Telegram orders bot knowledge base from product sheets, FAQs, or order instructions stored in Google Drive.
• Automatically keep your RAG dataset up to date whenever a team uploads new documentation.
• Enable SaaS operators to centralize knowledge in Google Drive while powering semantic search in Supabase.

Technical details

• Google Drive trigger: monitors a folder by ID and runs every minute.
• Extract from file: pulls plain text content from the downloaded document.
• LangChain OpenAI embeddings node (n8nn8n-nodes-langchainembeddings open ai): generates embeddings for text chunks.
• LangChain Supabase Vector Store node (n8nn8n-nodes-langchainvector store supabase): writes chunked documents and vectors to Supabase.
• Setup: provide Google Drive credentials, an OpenAI API key for embeddings, and Supabase connection details for the vector table.