chatgpt

Claude and ChatGPT’s New MCP Integration Features Are Insane

Dave Falcon

06 Jun 2025 • 4 min read

'Connectors’ in ChatGPT and ‘Integrations’ in Claude let you connect external tools and automate tasks easily

I think you heard about ‘connectors’ in ChatGPT!

No?

Then let me tell you.

Just yesterday (4th June), ChatGPT launched a new feature (tool) called ‘Connectors.’

With this feature, you can now integrate ChatGPT with any external tool or access any of your external data.

This happens with MCP integrations.

What does this mean?

It means you can now give ChatGPT access to tools or apps like Gmail, calendar, Google Drive, Hubspot, or any other app that you want.

ChatGPT can read, write, and publish any content on those apps.

By the end of this post, you will learn how you could set up connectors, what features you will get, and how helpful it will be for you.

How to Enable ChatGPT Connectors

To enable Connectors, go to ChatGPT.

Next, go to Settings.

You will find the Connectors right there.

Once you click on it, you’ll see tools like Dropbox, GitHub, Gmail, Google Calendar, Google Drive, HubSpot, and others.

ChatGPT settings showing enabled connectors like GitHub, Gmail, and Google Drive, with additional options to browse connectors such as Box, Dropbox, Google Calendar, and HubSpot.

Create Your Custom Connector (Beta)

Also, if you click “create” there, you’ll see you can add a new connector.

Now, it says this is a beta thing, mainly for developers. Basically, they’re just being extra careful about what you hook up to with MCP, ’cause you could connect to something you shouldn’t.

So, if you use this, please, please make sure that the MCP server is from a good company, a source you trust, not some iffy thing.

ChatGPT settings showing available connectors like Gmail, Google Drive, GitHub, and Outlook.

You need to confidently tick the “I trust this application” checkbox. Remember, OpenAI doesn’t check these custom connectors. So, they’re just making sure they’re not on the hook.

I want you to read this guide on ‘MCP Servers’. It’ll show you exactly how to set up an MCP server, build one, and all that stuff.

Form to create a custom connector in ChatGPT, including fields for name, description, MCP URL, and authentication.

Let me show you a real example of how to connect an external tool with ChatGPT.

Example: Connect ChatGPT with Gmail

We will connect Gmail with ChatGPT in this example. Once we click on the Gmail icon, we can see:

What it works with
What is the category
Who is the developer
What is the website
The privacy policy

They are very serious about getting you to connect MCP to things that you actually trust.

They say that permissions are always respected, private, and secure, and you’re in control of your data.

Next, continue with your Google account. Give it the necessary permission so it can read, compose, and send your emails.

ChatGPT settings showing Gmail connector details, including connection date, usage category, and developer info

Once it is connected, you will see that you can use it now for Deep Research.

Now let’s try this in ChatGPT.

Using Gmail Integration for Research

Open a new chat in ChatGPT and go with Deep Research. Then, from the sources, turn on Gmail.

Now, ChatGPT can pull any data from your emails: read emails, compose emails, and send emails.

For example, I asked about a random email to see what information ChatGPT tells me about it.

I asked ChatGPT, “Please write me a detailed report about the emails from Google Search Console.”

Since I had emails from Search Console, it asked if I was looking for indexing issues, search performance or something else.

ChatGPT asking the user clarification questions to tailor a detailed report based on emails from Google Search Console.

I just said, “I want to what exactly they informed me in the emails.”

I noticed it’s running the report for a long time. So I thought I needed to wait a while to see ChatGPT in action after its research was done.

ChatGPT reading emails from Google Search Console to generate a summary report, showing progress bar and sources.

Interestingly, Claude has launched the same feature just a day ago.

So I thought of trying it out.

When I tried both ChatGPT and Claude MCP connections, I found that one of them is working well as per its promise.

So let’s take a look at it.

ChatGPT vs. Claude: Who is the MCP Boss?

After trying both AI, I think Claude is far better than ChatGPT in MCP integrations.

In Claude, you can connect Gmail, Calendar, Drive, or even use Zapier for custom MCP integrations.

Claude settings showing active integrations with Gmail and Zapier, with options to connect GitHub, Google Drive, and Google Calendar.

For example, I used Claude for a similar task.

I connected it with Gmail and asked it to summarize the last email I received from James Clear. It acutally did the job and I did not have to wait for long as I waited for ChatGPT.

Claude interface showing a summary of the latest James Clear email using Gmail integration.

Now, if I talk about ChatGPT, I did “Deep Research” and then turned on Gmail. Even then, it can’t access my Gmail directly. It just tells me to copy and paste my emails. So, even if I give access in settings, it only works inside Deep Research.

Meanwhile, Claude can read emails, give summaries, and even write replies and send them. That makes it more useful in this case.

I would always choose Claude over ChatGPT in this scenario.

It has fewer built-in tools like ChatGPT, but if you set up your own custom MCP tools with Zapier, you can literally do anything you want. You can even do in-depth research like ChatGPT.

ChatGPT MCP integrations are very limited now. It doesn’t work as promised. It takes a lot of time to read an email, and the biggest thing is that it can’t even reply.

So I think ChatGPT is limited in terms of integrations at the moment, but Claude is crushing it.

Learn how to make money online using ChatGPT and other tools by creating animations.