How to Automate Daily Email Summaries Using Make.com and ChatGPT: A Step-by-Step Guide

Category: AI & Tech Tutorials | Published on: Taazamind.com

Introduction

If your inbox feels like a second job, you’re not alone. Professionals and content creators spend an average of 28% of their workday just reading and responding to emails — most of which could be summarized in a few bullet points. That’s exactly the problem Make.com (formerly Integromat) and ChatGPT solve together, when set up correctly.

The frustration usually comes from three places: not knowing which automation tool handles email best, struggling to pass structured data from Gmail or Outlook into an AI model, and getting ChatGPT responses that are too vague to be useful. Each of these is a configuration problem, not a platform limitation.

This guide walks you through a fully tested, end-to-end workflow — from connecting your email account in Make.com to generating clean, actionable daily summaries via ChatGPT’s API. Whether you’re a solo founder drowning in newsletters or a team lead managing client threads, every step here is written to be hands-on, logical, and replicable without a developer background.

Technical Specifications

Technical Detail	Specification / Requirement
Target Platform	Make.com (Free or Core plan) + OpenAI ChatGPT API
Compatible Email Providers	Gmail, Outlook / Microsoft 365, IMAP-supported providers
Difficulty Level	Beginner–Intermediate (No coding required)
Estimated Setup Time	45–75 minutes
Trigger Type	Scheduled (Daily Cron) or Webhook
AI Model Used	GPT-4o or GPT-3.5-Turbo (via OpenAI HTTP module)
Output Format	Plain text digest, email, Slack message, or Notion page
API Cost Estimate	~$0.002–$0.01 per daily summary (GPT-3.5)
Make.com Operations Used	5–10 ops per run (varies by email volume)
Prerequisites	Make.com account, OpenAI API key, Gmail/Outlook access

Step-by-Step Methods

Method 1: Connect Your Email to Make.com Using a Watch Module

This is your foundation step. Make.com’s “Watch Emails” module polls your inbox at a scheduled interval and feeds individual messages into the automation pipeline as structured data objects.

1. Log in to your Make.com account at make.com and click Create a new scenario.
2. Click the large “+” button in the scenario editor to add your first module.
3. Search for “Gmail” (or “Microsoft 365 Email” if you use Outlook) and select the Watch Emails trigger.
4. Authenticate your Google account when prompted — Make.com will request read-only access to your Gmail.
5. Configure the trigger settings: Set the folder to Inbox, the maximum number of emails per cycle to 50, and mark emails as Read after fetch to avoid re-processing.
6. Set a schedule: Click the clock icon at the bottom of the module and choose Every day at 7:00 AM (or your preferred time).

Why this matters: The Watch module doesn’t pull all your email at once — it processes only emails received since the last run. This keeps your operation count low and your summary focused on today’s messages only.

Method 2: Aggregate Multiple Emails Into a Single Text Block

ChatGPT works best when you send it one consolidated block of text, not 30 separate API calls. Make.com’s Text Aggregator module is built exactly for this.

1. Add a new module after your Gmail Watch module by clicking the small “+” connector circle.
2. Search for “Text Aggregator” under Make.com’s built-in tools and select it.
3. Set the Source Module to your Gmail Watch Emails module (this tells it which data to collect).
4. In the Text field, use the dynamic variables panel to build your template. A good format is: From: {{1.from}} Subject: {{1.subject}} Date: {{1.date}} Body: {{1.text}} ---
5. Set the Row separator to a new line (\n) so each email is clearly divided.
6. Leave “Group by” empty unless you want to group by sender domain (an advanced optional step).

Why this matters: Without aggregation, each email triggers a separate ChatGPT call — multiplying your API costs and making the summary incoherent. Aggregation hands GPT the full picture in one prompt.

Method 3: Send the Aggregated Text to ChatGPT via OpenAI HTTP Module

Make.com has a native OpenAI module, but using the HTTP module with the OpenAI API directly gives you more control over your prompt structure and model selection.

1. Add a new module and search for “HTTP” — select Make a Request.
2. Set the URL to: https://api.openai.com/v1/chat/completions
3. Set the Method to POST.
4. Click “Add Header” and enter:
   • Name: Authorization
   • Value: Bearer YOUR_OPENAI_API_KEY (replace with your actual key from platform.openai.com)
5. Set Body Type to Raw and Content Type to application/json.
6. Paste the following JSON into the body field, inserting your aggregated text using Make.com’s dynamic variable {{2.text}}:

{
  "model": "gpt-4o",
  "messages": [
    {
      "role": "system",
      "content": "You are an executive assistant. Summarize the following emails into a concise daily digest. Use bullet points. Highlight any action items, deadlines, or urgent messages at the top."
    },
    {
      "role": "user",
      "content": "{{2.text}}"
    }
  ],
  "max_tokens": 800,
  "temperature": 0.3
}

7. Click OK to save the module.

Why temperature 0.3? Lower temperature values make GPT’s output more focused and factual — exactly what you want for a business summary. Higher values (0.7+) produce more creative, less predictable output.

Method 4: Parse the ChatGPT Response and Send the Summary

The OpenAI API returns a JSON object. You need to extract just the summary text from it before sending it anywhere useful.

1. Add a new module and search for “JSON” — select Parse JSON.
2. Set the JSON String to {{3.data}} (the raw output from your HTTP module).
3. After the JSON parser, add another module — choose your delivery method:
   • Gmail → Send an Email: to receive the summary in your inbox each morning
   • Slack → Create a Message: to post it to a private Slack channel
   • Notion → Create a Database Item: to log it as a daily note
4. For Gmail delivery, configure the Send Email module:
   • To: Your email address
   • Subject: Daily Email Digest – {{formatDate(now; "DD MMM YYYY")}}
   • Content: Map the field {{4.choices[].message.content}} as the email body
5. Click Run Once to test the full scenario with real data.

Method 5: Add a Filter to Exclude Newsletters and Promotional Emails

Without filtering, your summary will be diluted with Zomato offers and Medium digests. A simple Make.com filter fixes this in under two minutes.

1. Click the small filter icon (the funnel-shaped icon) on the connector line between your Gmail Watch module and the Text Aggregator.
2. Add Condition 1: From → Does not contain → noreply
3. Click “Add AND rule” and set Condition 2: Labels → Does not contain → Promotions
4. Add a third condition if needed: Subject → Does not contain → unsubscribe
5. Click OK — the filter will now silently skip promotional emails before they enter the aggregator.

Pro tip: In Gmail, pre-label important senders with a custom label like “Priority” and add a Make.com filter to only include emails with that label. This gives you a VIP-only daily digest instead of a catch-all summary.

Frequently Asked Questions

Can I use this automation with a non-Gmail email account like Yahoo or a custom domain?

Yes — Make.com supports IMAP connections, which work with virtually any email provider including Yahoo Mail, Zoho Mail, and custom domain emails hosted on Hostinger, Namecheap, or cPanel servers. Instead of choosing the Gmail module, select Email → Watch Emails (IMAP). You’ll need your IMAP server address (e.g., imap.zoho.com), port (993 for SSL), and your email login credentials. The rest of the workflow — aggregation, ChatGPT API call, and delivery — remains identical. Note that Gmail and Outlook modules use OAuth authentication (more secure), while IMAP uses a password, so consider using an app-specific password if your provider supports it.

How do I stop ChatGPT from summarizing the same emails twice if the scenario runs again?

Make.com’s Watch Email trigger is stateful by default — it stores a “high water mark” timestamp after each run and only fetches emails received after that point. As long as you don’t reset the module manually, you won’t get duplicates. To verify this, click on the Gmail Watch module, scroll down, and confirm the Starting Point is set to From now on (not All emails). If you’re seeing duplicates, the most likely cause is that the scenario was manually run with “Run Once” in test mode, which bypasses the stateful cursor. For production, always activate the scenario using the toggle switch, not the Run Once button.

Is there a way to make the ChatGPT summary sound more like my personal writing style?

Absolutely. The system prompt in your OpenAI API call is the single most powerful lever for controlling output style. Instead of a generic instruction, write your system prompt in first-person and describe your preferences explicitly. For example: “You write like a senior product manager. Use plain English, avoid corporate jargon, and keep bullet points to one line each. Always flag emails from my boss or clients as HIGH PRIORITY at the top of the summary.” You can also add a few examples of good summaries directly in the system prompt (this is called few-shot prompting) to significantly improve consistency across days.

Final Thoughts

This workflow takes about an hour to set up and saves you 20–30 minutes every single morning. The real power here isn’t just the time saved — it’s the mental clarity that comes from starting your day with a curated digest instead of an overwhelming inbox. Once this is running, you can extend it further: summarize by sender, tag action items into a task manager, or even trigger a Slack standup update automatically.

If you run into issues with the HTTP module or OpenAI API responses, double-check that your API key has billing enabled and that you’re mapping the correct JSON path (choices[0].message.content) in the Parse JSON module. That’s the most common point of failure for first-time builders.

For more no-code AI automation guides built for real-world workflows, visit Taazamind.com.

Article by Taazamind.com — AI & Tech Tutorials written from hands-on experience.