ChatGPT to Integrate with Real-Time Tools Using MCP

In a recent tweet spotted by AI researcher Tibor Blahy and discussed widely across developer circles, OpenAI appears to be quietly testing Model Context Protocol (MCP) integration within ChatGPT’s web app.

https://x.com/btibor91/status/1922431656067268876

Toggle

Free SEO Audit: Uncover Hidden SEO Opportunities Before Your Competitors Do

Gain early access to a tailored SEO audit that reveals untapped SEO opportunities and gaps in your website.

Analyze My Site

ChatGPT is about to become more than a conversational AI and it is turning into a real-time, connected assistant that pulls live data from your favorite tools and custom apps.

According to McKinsey it estimates that companies could increase productivity by up to 40% if AI could plug directly into internal workflows.

This addition, now partially visible in some builds of the ChatGPT web app under a “Connectors” settings panel which suggests a bold step forward for how AI can tap into external data right from the tool you used.

What Is MCP and Why Should You Care?

MCP or Model Context Protocol is an open source specification that allows external systems to provide context to AI models during interaction.

 Think of it as a universal adapter. With it your company’s CRM, scheduling system or document repository can feed real-time data into ChatGPT which makes smarter decisions, automates tasks and 3x personalize outputs. 

MCP Support Being Tested Inside ChatGPT

OpenAI is rolling out internal tests for MCP which is an open-source protocol designed to allow third-party data and services to be accessed as context by AI models like ChatGPT. 

In simple terms you will soon be able to plug your favorite (or critical) tools such as Gmail, Google Sheets, Salesforce, Notion or even proprietary business apps directly into ChatGPT.

Users first spotted a new setting called “Connectors” in the ChatGPT web UI which includes an “Add Connector” option. This lets users enter the name, URL and description of a custom tool. These are the early signs that MCP is being embedded directly into the ChatGPT core experience rather than existing as a separate plugin system.

Now, pause for a second this sounds familiar, doesn’t it?

A Reboot of Plugins? Or Something Bigger?

https://x.com/btibor91/status/1922431656067268876

John Rehberger, again a notable voice in the AI community, commented below Tibor Blahy X post. 

“Oh, very interesting. Seems like a reboot of the plugins approach in main chat experience?”

Indeed, back in 2023 OpenAI launched plugins for ChatGPT, allowing limited integration with third-party tools like Expedia, Zapier and Code Interpreter. But the experience felt fragmented and the access was limited to paid tiers and developers had to jump through hoops to deploy and maintain a plugin. 

Enter MCP: an open, flexible standard that removes that friction.

Developers can now expose APIs and services via MCP and ChatGPT can tap into them using what seems to be a new kind of “AI Action.” According to insiders, this might soon appear alongside existing options like OpenAPI when creating Custom GPTs. 

For Enterprises, the Real Value Lies in Customization

Now, when it comes to business it really makes a difference. 

Imagine you run a fintech startup. You use PostgreSQL for your transactional data, Stripe for payments and Intercom for customer service. MCP will let you build a unified assistant that understands customer context from all three automatically responds to finance-related questions and even flags anomalies all within a single chat interface.

This is about contextual intelligence which has long been the holy grail of AI-powered productivity.

• Automating internal ticket triage by pulling data from Jira, GitHub and Slack. 
• Generating real-time executive dashboards using company-specific KPIs.
• Sales prospecting GPTs that use internal win-loss data alongside tools like Apollo or LinkedIn Sales Navigator.

What’s Next? When Will MCP Be Live?

As of now, MCP is in internal testing. No official release date has been announced but given the visibility in the UI and growing chatter among AI developers it can be expected that a formal rollout could be just weeks away.

How MCP servers will be supported with ChatGPT? Can we guess? – Feature requests – OpenAI Developer Community

Some predict that OpenAI may launch this as part of its next Dev Day or major release, perhaps bundled with broader announcements around the future of Custom GPTs, AI Actions and Team/Enterprises plans. 

Once available, expect developers and SaaS platforms to begin offering prebuilt MCP connectors similar to what Zapier did for plugins. This could create an ecosystem of AI-powered workflows tailored to verticals like healthcare, finance, logistics and education. 

Overall, we have long known that context is the missing piece in generative AI and now with MCP we are getting closer to solving it. If this rolls out as expected, ChatGPT won’t just talk. It will understand. It will access. And it will act within the systems you already use every day. 

The future of AI is no longer just intelligent. It is integrated