Managing inboxes manually can be overwhelming, especially with high email volumes. Automating this process with Make and ChatGPT can save time and streamline your communications by automatically organizing emails and sending context-specific AI responses.

In this tutorial, you’ll learn how to build a comprehensive inbox Manager and AI Auto-Reply Assistant, covering all necessary tools and configurations.

Key Objectives

1. Automate Inbox Management: Use Make to automatically filter and organize incoming emails based on specific criteria.
2. Leverage AI for Auto-Responses: Integrate ChatGPT with Make to generate context-aware replies to emails.
3. Enhance Efficiency: Reduce manual intervention by automating responses to common inquiries and categorizing emails for further action.
4. Personalize and Customize Workflows: Use filters and AI prompts to tailor responses based on the type of email received.
5. Implement Advanced Options: Incorporate follow-ups, customize email templates, and refine workflows for optimal results.

Set Up the Inbox AI Assistant

Step 1: Setting up the Make scenario

The first step is to create an account on Make on log in, navigate to the dashboard, and create a New Scenario.

Step 2: Adding your email provider module

In your scenario, search for your email provider (e.g., Gmail, Outlook) as the trigger.

Select the “Watch Emails” action, which monitors your inbox for new incoming emails. Also, use this tutorial to properly connect Gmail to Make. It's well explained and should take 5 minutes to do.

Set the trigger to check for new emails in your Inbox folder. Use criteria "All emails" so that it also classifies emails that you've already opened.

If you want Gmail to watch the emails you've already received and categorize them, increase the Maximum number of results to 10 for instance

Finally, before we work on the next step of the automation, make sure you have an email that you can use as a test in your inbox. If not, just send an email to yourself.

Step 3: Setup a filter for emails to check for specific keywords

The third step is to set filters to manage emails more efficiently. For example, set up rules to filter emails that contain specific keywords like "Urgent", "Inquiry", or "Meeting Request." The keywords should be common and based on the specific emails you want to watch for.

Example filter: 

Condition: Email Subject contains “Meeting Request.”

Make will only trigger responses for emails with this subject.

Integrate ChatGPT for Auto-Responses

Step 1: Set Up ChatGPT Module in Make

Now that Make is set to trigger when an email arrives, we need to integrate it with OpenAI's API to categorize the email and write the reply that will be used.

Create a new module, type "OpenAI" and select Create a Completion. Make sure to connect OpenAI to your make account using your OpenAI API key (more about this here).

Select the model gpt-4o-mini, which is powerful and cheap enough for the use case.

Step 2: Map the Email Content to the ChatGPT Prompt

The second step is to use Make’s mapping feature to map the email body from the Email trigger Module into the ChatGPT Module.The most important part of this whole setup is the prompt. It determines the assistant behaviour.

Prompt:

You are an AI email auto-responder for {{name}}.
Your role is to handle specific types of emails with predefined responses. Follow the guidelines below to ensure accurate and professional responses. If you're uncertain, provide a polite fallback response.
Email Body: [email_body]
Sender Name: [sender_name]
Email Subject: [email_subject]
Guidelines:
Clear Boundaries: Only respond to meeting-related emails, product inquiries, or requests for general information.
For all other emails, respond with: 'Thank you for your message. The team will follow up soon.'
Fallback for Insufficient Info: If you don’t have enough information to provide a detailed answer, respond with: 'I don’t have all the details right now, but someone will reach out shortly.
Specific Triggers: Only reply to emails that contain key terms like 'meeting,' 'inquiry,' or 'request.'
Pre-built Templates:
- Meeting Request: 'Thank you for your message! I’ve received your meeting request and will confirm the time shortly.'
- Product Inquiry: 'Thank you for your interest in our products. We’ll send detailed information soon.'
- Sentiment Awareness: If the email expresses frustration or negative sentiment, avoid providing a direct response and instead reply with: 'We understand your concern and are escalating this to the relevant team.'
Use these responses to ensure professional, safe communication, and always prioritize deferring to a human when uncertain.
Make sure to return just the response generated, email subject for reply to be sent, if the email received the email type we want to respond to or not, and if human intervention is needed in a json format

Insert this prompt and map in the email subject, body and sender name. Don't forget to give the assistant as much information as possible!

Then, select "Show advanced settings" and modify the Temperature to around 0.7. This will ensure the model will be more creative with it’s replies and not just give generic answers.

Make sure to return just the response generated, email subject for reply to be sent, if the email received the email type we want to respond to or not, and if human intervention is needed in a json format

In the ChatGPT Module, set the Response Format to Json Object and Set the Parse Json to Yes. The AI will give a mappable json output directly.

Set Up the Auto-Reply System

Step 1: Send the generated response via email provider module

Add another module in Make for sending emails using your email provider (e.g., Gmail, Outlook).

• Configure the "Send an Email" action.
• Set the recipient to the original sender's email.
• Map the ChatGPT response as the body of the email and subject as subject of the reply email.

Step 2: Test and refine the scenario

The second step here is to test the scenario by sending a test email with the subject “Meeting Request” and check if the workflow triggers ChatGPT to generate a professional reply.

Refine the AI prompts and filters if responses are too generic or missing key details.

It worked!

‍Advanced option and enhancement ideas

 1. Handling Different Email Types

   - Create more specific workflows for emails containing words like “urgent,” “pricing,” or “support request.”

   - Customize ChatGPT prompts to reply with specific details for each type.

Example Use Case: Product Inquiry Auto-Reply

1. Email Received: A customer inquires about your products via email.

2. ChatGPT Response: ChatGPT auto-generates a polite, informative reply:

   - Response: “Thank you for your inquiry! We will send more details about our products and pricing shortly.”

Conclusion

Congratulations on successfully setting up your automated Inbox Manager and AI Auto-Reply Assistant using Make and ChatGPT!

This powerful system will streamline your email management, saving you time and enhancing your productivity by automatically filtering emails and generating context-aware responses.