Real-time Email RAG Assistant with Gmail, OpenAI GPT, and PGVector

This n8n workflow creates a real-time RAG (Retrieval Augmented Generation) assistant for your emails. It automatically processes incoming emails, extracts their content, and stores it in a PGVector database. When you interact with the AI assistant, it retrieves relevant email information from the database to provide contextually rich and accurate responses.

What it does

This workflow automates the following steps:

1. Monitors Gmail: It continuously listens for new emails in your specified Gmail account.
2. Loads Email Content: For each new email, it extracts the full content.
3. Splits Text: The email content is then broken down into smaller, manageable chunks using a Recursive Character Text Splitter. This is crucial for efficient vector embedding and retrieval.
4. Generates Embeddings: OpenAI's embedding model converts these text chunks into numerical vector representations.
5. Stores in PGVector: The generated embeddings, along with the original email text, are stored in a PostgreSQL database with the PGVector extension. This creates a searchable knowledge base of your emails.
6. AI Assistant Interaction: When a chat message is received (e.g., from a user querying the assistant), an AI Agent is triggered.
7. Retrieves Relevant Information: The AI Agent queries the PGVector store to find the most relevant email content based on the user's chat message.
8. Generates Response: Using the retrieved email content as context, an OpenAI Chat Model generates a comprehensive and relevant response to the user's query.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance.
• Gmail Account: Configured as a credential in n8n for the Gmail Trigger.
• OpenAI API Key: Configured as a credential in n8n for the Embeddings OpenAI and OpenAI Chat Model nodes.
• PostgreSQL Database with PGVector Extension: A PostgreSQL database with the pgvector extension enabled. You will need the connection details (host, port, database, user, password) configured as a credential in n8n for the Postgres PGVector Store node.
• LangChain Nodes: Ensure the @n8n/n8n-nodes-langchain package is installed in your n8n instance.

Setup/Usage

1. Import the Workflow: Download the JSON provided and import it into your n8n instance.
2. Configure Credentials:
   • Gmail: Set up a new Google OAuth2 credential for Gmail.
   • OpenAI: Set up a new OpenAI API Key credential.
   • PostgreSQL: Set up a new PostgreSQL credential with the details for your PGVector-enabled database.
3. Activate the Workflow: Once all credentials are configured, activate the workflow.

The workflow will now automatically process new emails and make their content available for your AI assistant to query. You can then interact with the AI assistant via the "When chat message received" trigger (e.g., through a custom chat interface or another n8n workflow that sends messages to this trigger).