n8n Workflow: Supabase Vector Store RAG Chatbot

This n8n workflow demonstrates a Retrieval-Augmented Generation (RAG) pattern using Supabase as a vector store to answer questions based on external documents, triggered by a chat message. It integrates OpenAI for embeddings and chat model capabilities.

Description

This workflow simplifies the process of building a RAG-powered chatbot. It allows you to ingest documents (e.g., from Google Drive), process them, store their embeddings in a Supabase vector database, and then use that knowledge base to answer user questions received via a chat interface.

What it does

1. Triggers on Chat Message: The workflow starts when a chat message is received, acting as the user's query.
2. Loads Documents: It retrieves documents from a specified Google Drive folder using a Default Data Loader.
3. Splits Text: The retrieved documents are then split into smaller, manageable chunks using a Recursive Character Text Splitter to optimize for embedding and retrieval.
4. Generates Embeddings: OpenAI Embeddings are generated for these text chunks.
5. Stores in Supabase Vector Store: The text chunks and their corresponding embeddings are stored in a Supabase vector database.
6. Retrieves Relevant Information: For a given chat query, the workflow retrieves the most relevant document chunks from the Supabase Vector Store.
7. Answers Questions: It uses an OpenAI Chat Model in conjunction with the retrieved information to generate a comprehensive answer to the user's question via a Question and Answer Chain.
8. Prepares Data (Optional/Placeholder): An "Edit Fields" node is present, likely for data manipulation or transformation, though its specific configuration is not detailed in the provided JSON.

Prerequisites/Requirements

• n8n Account/Instance: To run the workflow.
• Supabase Account: With a configured vector database. You will need your Supabase URL and API Key.
• OpenAI Account: For generating embeddings and using the chat model. You will need an OpenAI API Key.
• Google Drive Account: If you intend to use the Google Drive document loading functionality. You will need Google Drive credentials configured in n8n.

Setup/Usage

1. Import the Workflow:
   • Download the provided JSON file.
   • In your n8n instance, go to "Workflows" and click "New".
   • Click the "Import from JSON" button and paste the workflow JSON or upload the file.
2. Configure Credentials:
   • Locate the "Supabase" node and configure your Supabase credentials (URL and API Key).
   • Locate the "Embeddings OpenAI" and "OpenAI Chat Model" nodes and configure your OpenAI API Key credentials.
   • If using Google Drive, configure your "Google Drive" node with appropriate credentials.
3. Configure Nodes:
   • Google Drive: Specify the folder or files you wish to load documents from.
   • Recursive Character Text Splitter: Adjust chunk size and overlap as needed for your documents.
   • Supabase Vector Store: Ensure the table name and other settings match your Supabase vector database configuration.
   • Question and Answer Chain: Review the prompt and model settings.
   • Chat Trigger: Configure how your chat messages will be received (e.g., via a specific chat platform integration if you extend the workflow).
4. Activate the Workflow: Once configured, activate the workflow to start listening for chat messages.
5. Test: Send a chat message to the configured trigger to test the RAG functionality.