This document outlines the architecture, setup, and usage of a Make.com scenario that automates the summarization of customer feedback submitted via a Google Form, using Hugging Face’s Inference API, and sends the summarized feedback via Gmail. The solution is designed to operate within the free tiers of Make.com, Hugging Face, and Gmail.

Solution Overview

The workflow automates the process of collecting feedback, summarizing it with AI, and emailing the results. It integrates:

  • Google Forms: Captures customer feedback.
  • Hugging Face Inference API: Summarizes feedback using the facebook/bart-large-cnn model.
  • Gmail: Sends an email with the original feedback and its summary.
  • Make.com: Orchestrates the automation.

Objectives

  • Detect new feedback submissions in real-time.
  • Summarize feedback using a free AI model.
  • Email the original feedback and summary to a specified recipient.
  • Operate within free tier limits to minimize costs.

Architecture Diagram

Setup Instructions

Step 1: Configure Google Form

  1. Go to Google Forms and create a form titled “Customer Feedback.”
  2. Add two questions:
    • Question 1: “Name” (Short answer text).
    • Question 2: “Please share your feedback about our service” (Paragraph).
  3. Save the form and test by submitting a sample response (e.g., Name: “John Doe,” Feedback: “The service was great, but the wait time was too long.”).

Step 2: Set Up Make.com Scenario

  1. Log in to Make.com and create a new scenario named “Feedback Summarization Workflow.”
  2. Click the “+” button to add the first module.

Step 3: Add Google Forms Trigger

  1. Module: Google Forms – Watch Responses
  2. Configuration:
    • Connect your Google account.
    • Select the “Customer Feedback” form.
    • Set the trigger to detect new responses.
  3. Test: Submit a test response to confirm the module captures the name (e.g., {{1.2}}) and feedback (e.g., {{1.1}}).

Step 4: Extract Feedback Text

  1. Module: Google Forms – Get a Response
  2. Configuration:
    • Connect your Google account.
    • Select the “Customer Feedback” form.
    • Map the response ID from the previous module (e.g., {{1.responseId}}).
    • Select the fields: “Name” and “Please share your feedback about our service.”
  3. Output: Name (e.g., “John Doe”) and feedback text (e.g., “The service was great, but the wait time was too long.”).

Step 5: Summarize Feedback with Hugging Face

  1. Module: HTTP – Make a Request
  2. Configuration:
    • URL: https://api-inference.huggingface.co/models/facebook/bart-large-cnn
    • Method: POST
    • Headers:
      • Authorization: Bearer {{your_hugging_face_api_token}} (replace with your token from Hugging Face).
      • Content-Type: application/json
    • Body Type: Raw
    • Content Type: JSON
    • Request Content:
    • {
    •   “inputs”: “{{2.feedback_text}}”,
    •   “parameters”: {
    •     “max_length”: 100,
    •     “min_length”: 30,
    •     “do_sample”: false
    •   }

}

  • Map {{2.feedback_text}} to the feedback text from Step 4 (e.g., {{2.1}}).
  1. Output: JSON response with the summary (e.g., [{“summary_text”: “Great service but long wait times.”}]).

Step 6: Parse Hugging Face Response

  1. Module: JSON – Parse JSON
  2. Configuration:
    • JSON String: {{3.body}} (HTTP response body from Step 5).
  3. Output: Summary text (e.g., {{4.summary_text}}).

Step 7: Send Email with Gmail

  1. Module: Gmail – Send an Email
  2. Configuration:
    • Connect your Google account.
    • To: Your email address (e.g., yourname@gmail.com).
    • Subject: Feedback Summary of {{1.2}}
      • {{1.2}} maps to the name from the trigger (e.g., “John Doe”).
    • Content:
    • Name: {{1.2}}
    • Feedback: {{1.1}}

The summary of the Feedback text is: {{4.summary_text}}

  • {{1.2}}: Name from the trigger.
  • {{1.1}}: Feedback text from the trigger.
  • {{4.summary_text}}: Summary from Step 6.
  1. Output: An email sent to the specified address with the name, feedback, and summary.

Usage Instructions

  1. Submit Feedback:
    • Access the Google Form and submit a response (e.g., Name: “Jane Smith,” Feedback: “Friendly staff, but the product was overpriced.”).
  2. Scenario Execution:
    • The scenario runs automatically when a new response is detected (if active).
    • Alternatively, use Make.com’s “Run Once” button to test manually.
  3. Check Email:
    • Verify that an email arrives with:
      • Subject: “Feedback Summary of Jane Smith”
      • Body:

Name: Jane Smith

Feedback: Friendly staff, but the product was overpriced.

The summary of the Feedback text is: Friendly staff but overpriced product.

Conclusion

This Make.com scenario provides an efficient, cost-free solution for summarizing customer feedback and emailing the results. By leveraging Google Forms, Hugging Face’s Inference API, and Gmail, it automates a valuable workflow while staying within free tier constraints. For scalability, monitor usage and consider paid plans if limits are exceeded.

RELATED ARTICLESMORE FROM AUTHOR