Feb 13, 2025 | 7 minutes
Make and MindStudio - a solution to email overload

Remember the days of drowning in a sea of emails, desperately trying to sort through newsletters, marketing blasts, and urgent requests? Those days are over.Â
This guide will walk you through using AI to automatically label and organize all incoming emails. This is important for every professional receiving hundreds of emails per day to keep their sanity and only reply to the ones that matter. AI is incredibly accurate and will replace all deterministic logic in your inbox.
Introduction to AI-powered email labeling
AI email labeling solutions use artificial intelligence to organize emails automatically using advanced models that go far beyond an âif/thenâ rule you can set up in Gmail. Think of it as a digital assistant that understands the content of your emails and sorts them into relevant categories.
The difference between AI labeling and basic filters that follow match patterns is the fuzzy logic. With AI models, you donât need to specify rules like âif the email contains âstripeâ then label it as âbillingââ. Large language models are capable of understanding context, have knowledge of the surrounding topics, and can properly assign your emails to one of your predefined labels without any real trigger or rule other than... asking.
This effectively means that no matter what and who sends you an email, AI will be able to read it, understand it, label it, and organize it for you in a few seconds. And it can do so programmatically, 24/7.
Building your AI email tagging workflow
To get started, youâll need a workflow that takes in all incoming emails, labels them, and outputs the label for each. This is what the organization actually looks like: automatic labels for your account.
You can fetch all emails and tag them using Make.com, but youâll need AI to create the logic for tagging. In this example, weâre going to use MindStudio.
MindStudio is a powerful low-code platform that allows you to build custom AI workflows without writing code. Think of it as your IDE whether youâre technical, non-technical, or a mix of the two.Â
When using MindStudio flows through Make.com, you get several advantages over using OpenAI's module directly:
Access to over 50 different AI models
The platform is model-agnostic, so you're not locked into one provider
It supports image, audio, and text generation
It includes web scrapers and integrations with platforms like X, LinkedIn, and Google Trends
Multi-step AI workflows, you can have 10+ steps in one and call only one operation from Make
You can access all these features by calling your MindStudio flow directly from Make.com. Itâs as easy as copying and pasting a couple of authorization tokens.
Letâs get started with the build.
Create the AI Agent in MindStudio
Sign up for free
You'll need an account to build any workflow in the platform. Signing up is free, and thereâs no need to pay for a subscription. The forever free plan comes with $5 one-time in AI compute credits (thatâs over 14,000 articles generated with cheap models like Gemini 1.5 Flash) and 10,000 free API runs per month.
Generate your first agent
To create your first agent, click on âCreate AI Agentâ and âOpen Blank Projectâ
MindStudio agents can run within the user interface or through API. Given the emails will be coming from Make modules, we need an agent that runs through API.Â
Note: if you donât want to set all this up manually, you can use the âGenerate Agentâ feature.
Set up the API variables
To initialize variables in the workflow, click on the âStartâ block, the first blue module in the automation canvas, and initialize the variables you need to label the emails. For example:Â
Sender: this is important to recognize if itâs an internal or external email
Date: to give AI context about urgency and deadlines
Subject Line: to understand the topics
Body: to finalize the labeling with complete understanding
Generate the labels
Itâs time to decide on labels. In our example, weâll use billing, internal, urgent, newsletter, and announcement.
You can have as many or as few as you like. The best part about building your AI solutions is that you get to decide what matters most to your business.
To uphold the labels, we need two components: a system message, which is the AIâs main prompt (aka the âpersonalityâ or scope of the tool), and a Generate Text block, the actual prompt to generate the label. Letâs start with the main prompt. Something along the lines of:Â
âAssistant is an email inbox organizer.Â
It takes in emails from the user's Gmail inbox (passed via API) and labels them according to the user requirements.
Assistant does not assist in anything unrelated to email labeling, and strives to be truthfulâ
This should be all we need. Hereâs how it would look like in the tool:
Then, letâs go back to the automation canvas and add the labeling block. Click on the â+â icon and select the âGenerate textâ block. Then, write a prompt for the labeling process. In our example:
âGiven this information about the email:
Sender: {{sender}}
Date: {{date}}
Current Date: {{currentDate}}
Subject Line: {{subjectLine}}
Body: {{body}}
â
Assistant is tasked to label the email in only **ONE** of the following labels:Â
*Billing
*Internal
*Urgent
*Newsletter
*Announcement
Assistant will reply with the label and nothing elseâ
Weâre almost done. Now, we can decide to either display this result to the user (and the API), or save it as a variable.Â
You should display to the user if ALL you want from this workflow is the labeling. However, if you want to do something else, you should save it as a variable to reference it later on.
Letâs save it as a variable named âlabelâ.
AndâŠthatâs just about it. Just to show you the potential of multi-blocks flows, though, letâs add a âGenerate Textâ block to output the priority from 0-5 of the email content. You can rename any block in MindStudio and even add notes to remind yourself of what specific sections do.
Letâs move on to the final step: the end block.
Structure your output
Every text generation in MindStudio can output text (default), JSON, or CSV. Some blocks, like the âGenerate Assetâ block or function blocks for third-party integrations, can also output JavaScript / Python code or HTML templates. In our example, we simply have two text strings, each in its own âcontainerâ, the variable.
Label is in âlabelâ, which we can reference with => {{label}}
Priority is in âpriorityâ, which we can reference with => {{priority}}
What we want is an end block, instead of the chat block, that returns a JSON body with the values. Something like:
{
âLabelâ:âvalueâ
âPriorityâ:âvalueâ }
To do that, think about the structure, and plan it out in the JSON builder:
Now, whenever we call this flow, the output will include the JSON body, which we can use in Make to label the emails in Gmail.Â
Publish the agent, and youâre done.
Set up the Make automation
Once you have the email tagging system in MindStudio, you need to plug it into your Make.com scenario, which will look something like this:Â
Trigger whenever a new email enters a specified inbox, in this example a Gmail Inbox
When that happens, call the MindStudio AI Agent and send the sender, date, subject line, and body as variables in the MindStudio module
Parse the JSON to expose the two values: label and priority
Add a router and update the email according to the âlabelâ, e.g. if the label equals âurgentâ, then label it as urgent, if itâs ânewsletterâ, then label it as ânewsletterâ
The trigger
For the trigger, add a âWatch emailsâ for the Gmail module, select the folder youâd like to monitor (in our example, âMakeâ), and whether the email should switch to âreadâ once the AI processes it.
Running the MindStudio agent
When you publish a agent, opening it exposes all the values you need to make it run in Make. The platform even shows a step-by-step guided tutorial on how to connect the agent to your scenarios.
You will need an agent ID, to use the right agent, and the API key (tied to your workspace). You can also select a specific workflow within the AI agent, but we only have one for this example => âMainâ.
In Make.com, use the module and paste in the app ID/agent id, and API key (if you donât have a connection already), and initialize the variables specified in step one of this tutorial.
To make sure everything is running correctly, feel free to give it a test run or run the MindStudio module only with test variables.
Parsing the result
Now we have the result, but itâs within an unparsed collection. To use each value later on, we need to transform it into JSON within the scenario, parse the JSON, and then use the dynamic values from the parse in the next actions.
The result is a âParse JSONâ block containing one bundle with two values: label and priority.Â
Awesome â time to label the emails!
Tag the email
Finally, add a router that redirects to 5 different âModify Labelâ Gmail modules, one for each of the possible labels from the AI agent:Â
Billing
Internal
Urgent
Newsletter
Announcement
In our example, every filter looks for label = name (case insensitive).Â
Thatâs it: youâre done. All your incoming messages will now be orderly labeled and sorted for you automagically 24/7, and every labeling will cost less than a penny.
Resources
Short on time?Â
No problem. You can download the Make Template here and the MindStudio agent here. You need an account with both platforms to import the templates.
