Smart Customer Support System with GPT-4o, Gmail, Slack, and Drive Knowledge Base

This n8n workflow automates and enhances customer support by intelligently processing incoming emails, leveraging a knowledge base stored in Google Drive, and routing customer inquiries to the appropriate channels. It uses AI to classify emails, extract relevant information, and provide smart responses or actions.

What it does

This workflow streamlines customer support by:

1. Monitoring Incoming Emails: Automatically triggers when new emails arrive in Gmail.
2. Monitoring Google Drive for Knowledge Base Updates: Triggers when new files are added to a specified Google Drive folder, allowing for dynamic updates to the knowledge base.
3. Loading and Processing Knowledge Base Documents: When new files are added to Google Drive, it loads the document content, splits it into manageable chunks, and creates embeddings using OpenAI.
4. Storing Knowledge Base Embeddings: Stores these embeddings in Pinecone, creating a searchable vector store for AI queries.
5. Classifying Incoming Emails: Uses an AI Text Classifier (powered by OpenAI) to determine the category or intent of incoming customer emails.
6. Conditional Routing based on Classification: Routes emails based on their classification.
   • If classified as "Support":
     • An AI Agent (powered by OpenAI) processes the email query, potentially searching the Pinecone vector store (knowledge base) for answers.
     • It then performs an HTTP request, likely to a Slack channel or another internal system, to escalate or respond to the support query.
   • If classified as "Sales":
     • An AI Agent (powered by OpenAI) processes the email query, potentially searching the Pinecone vector store (knowledge base) for answers.
     • It then performs an HTTP request, likely to a CRM or sales team Slack channel, to handle the sales inquiry.
   • If classified as "Other":
     • An AI Agent (powered by OpenAI) processes the email query.
     • It then performs an HTTP request, likely to a general inbox or a different internal team, for manual review.
7. Logging to Google Sheets: Records details of the processed emails and their classifications in a Google Sheet for tracking and analysis.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance (cloud or self-hosted).
• Gmail Account: Connected as a credential in n8n for the Gmail Trigger.
• Google Drive Account: Connected as a credential in n8n for the Google Drive Trigger and Google Drive node.
• OpenAI API Key: For the AI Agent, OpenAI Chat Model, Embeddings OpenAI, and Text Classifier nodes.
• Pinecone Account: For the Pinecone Vector Store to manage your knowledge base embeddings.
• Google Sheets Account: Connected as a credential in n8n for the Google Sheets node.
• Slack Account / HTTP Endpoint: An HTTP endpoint or Slack webhook configured to receive messages from the workflow (for support, sales, and other inquiries).

Setup/Usage

1. Import the Workflow: Download the JSON provided and import it into your n8n instance.
2. Configure Credentials:
   • Set up your Gmail credential for the "Gmail Trigger" node.
   • Set up your Google Drive credential for the "Google Drive Trigger" and "Google Drive" nodes.
   • Set up your OpenAI API key credential for the "OpenAI Chat Model", "Embeddings OpenAI", "AI Agent", and "Text Classifier" nodes.
   • Set up your Pinecone credential for the "Pinecone Vector Store" node.
   • Set up your Google Sheets credential for the "Google Sheets" node.
   • Configure the HTTP Request node with the appropriate URL and authentication for your Slack webhook or other internal system.
3. Configure Nodes:
   • Gmail Trigger: Specify the mailbox and any filters for incoming emails.
   • Google Drive Trigger: Specify the folder ID in Google Drive where your knowledge base documents will be stored.
   • Pinecone Vector Store: Configure your Pinecone index name and environment.
   • Text Classifier: Define the categories for email classification (e.g., "Support", "Sales", "Other") and provide example texts for each category to train the classifier.
   • AI Agent: Customize the agent's instructions and tools if needed.
   • HTTP Request: Adjust the payload and URL for your specific Slack channels or other endpoints for different email classifications.
   • Google Sheets: Specify the Spreadsheet ID and sheet name where you want to log the data.
4. Activate the Workflow: Once all credentials and nodes are configured, activate the workflow.

The workflow will now automatically process new emails, update its knowledge base from Google Drive, classify inquiries using AI, and route them to the appropriate teams or systems, logging the details in Google Sheets.