✉ 🔊 A Daily Scenario to Generate an Audio Summary of Your Emails!


This showcase describes how to create a scenario in Make to generate an audio file of summaries of emails received in the past 24 hours, with screenshots of how each module is set up. The output of the scenario is an audio file that can be played on a computer or mobile device. A sample audio file is also provided if you want to hear how it could sound.

Modules Overview

You’ll need:

  1. your email client’s module
  2. a text aggregator to combine all the emails into a single variable, and replacing unnecessary characters in emails at the same time
  3. ChatGPT obviously :slight_smile:
  4. Google Text-to-speech, or something similar
  5. Google Drive, or somewhere that you can store the audio file and return the link in the output bundle


I used Gmail; you can also use the generic Email search module. Here are the settings:

You’ll need to set the maximum number of results high enough to catch all of the emails you regularly receive each day. If you have too many emails, I suggest setting a twice-daily (or more frequent) schedule.


Here’s where the text parsing happens. You’ll need to strip HTML, remove unnecessary spaces, urls, from the text. Gmail module allows you to select the email’s “Text Content” so I didn’t have to strip the HTML formatting. I limited the content of the email to the first 2000 characters as well.

You can also use a “HTML to text” module if your email trigger doesn’t have text output:


You can use any other AI you prefer, but this does the job.

I used the “gpt-3.5-turbo-16k-0613” model because that’s cheaper than gpt-4. Set the instructions in the System message, and the aggregated text in the User message. You need to also set the “Max Tokens” to a very high number so it can process all of your emails in one go.

Here’s my current System prompt, you can use this or tweak it to your preference:

You are a personal assistant handling incoming emails. Omit less-important emails such as spam, shopping deals, online meeting invites, privacy policy updates, one-time passwords and verification codes. Then, summarise the remaining emails in plain text format up to two sentences long. Do not include PII, emoji, or HTML. Begin each message by stating who it is from, what the subject is. Separate each email with two newlines.

Asking it to separate each email with three periods and two newlines so that the Text-to-speech module will generate a short pause in-between each.


I used Google Cloud Text-to-Speech here. If you have not set up any Google APIs before, there is a link under the connection field to a Help Article describing how to set it up.

File Storage

Google Drive lets me store the audio file and return the link in the output bundle.

Notify yourself; Listen-in!

From here you have a range of options on how you want to get the link to listen to the audio file.

I made it ping my desktop using the Integromat Customizable Google Chrome extension, as well as send a notification to my phone using the Make Android app.


Clicking on either will open the file in Google Drive, but only on Android does it auto-play.



Audio file example (I limited it to three emails and censored my email address):


The benefits of using this scenario include:

  • It saves you time by summarising your emails for you
  • It allows you to listen to your emails instead of reading them, which can be helpful if you are busy or on the go

The limitations of this scenario include:

  • It may not be able to summarise all of the information in an email, depending on the length and complexity of the email
  • The quality of the summary will depend on the quality of the ChatGPT model that you use

Possible improvements include:

  • Splitting up too many emails into several prompts for ChatGPT so it can handle it all (if you have 100s of emails daily)
  • Only process important emails, or specific labels; ignore mailing lists (spam - Gmail already does a good job of removing spam)
  • Also delete the previous audio file to save storage, or,
  • Organise the audio files into sub-folders for each month

Hey there @samliew :wave:

Thank you so much for sharing this valuable content with us :pray:.

This is so clever! I’m definitely gonna set this scenario up and enjoy my evening with a podcast made of my emails :smile:.

Keep up the great job :sunny:

1 Like

On potential refinement to this is to use Eden AI rather than Open AI and Google TTS directly.

Eden AI provides a single API for a whole range of AI services across multiple AI platforms. It means you can compare and contrast results from different platforms easily, and also means you only need one Eden AI subscription rather than multiple subscriptions.

The Eden AI app was recently launched on Make - give it a try!


(post deleted by author)

This is awesome!
How can I do something similar but with my google chat message groups? I have no time to read all of them.

(post deleted by author)