On Wednesday, OpenAI released a major update to ChatGPT that makes it much more useful at work—and offers a clear view of where tools like this are headed.
There are a number of new features (outlined below), but the most important is the introduction of “connectors”—which let ChatGPT pull in data from the tools your team already uses, like Outlook, Google Drive, SharePoint, HubSpot, and others.
Before getting into the specifics, it’s worth pausing to consider why this is such a meaningful shift.
If you’ve been in one of my keynote talks or workshops, you may have heard me explain the difference between general-purpose (horizontal) AIs and specialized (vertical) AIs.
General-purpose AIs—like ChatGPT—are designed to help with a wide range of tasks. They’re not tied to one function or tool. Over time, they’ll become the primary interface through which you manage your calendar, access files, follow up on deals, write summaries, or even coordinate projects—because they’ll understand what you need and know where to find it.
Specialized AIs, by contrast, are built to handle one system or domain—like a sales bot that only knows your CRM. They’re fast and focused, but siloed. They can’t pull in context from your documents, emails, or customer history.
For general-purpose AIs to be truly effective, they need to work across all the systems you do. That’s exactly what OpenAI’s new update enables. ChatGPT can now connect directly to the systems you rely on: pulling up a file from SharePoint, finding last week’s client emails in Outlook, summarizing a lead’s activity from HubSpot, or even querying your internal tools through a new protocol called MCP.
The experience is still in its early stages—it won’t replace your use of other systems yet. But marks the start of a major shift: AI not just as a chatbot, but as a unified interface for your work. And your time spent switching between tools may start to shrink. After all, why log into your CRM system if your AI can find the information for you?
OpenAI isn’t the only one racing in this direction. Anthropic recently announced similar features, and there’s a wave of companies adopting the MCP protocol to connect systems and even link AIs to one another. The foundation is being laid for how most of us will interact with information—and with work itself—in the years to come.
Now, onto the technical details of what is in this release.
The following capabilities are either newly released or expanded:
These connectors lay the foundation for making ChatGPT a central interface for business tools. Now, you can reference information across almost any system from within ChatGPT, and although still slow and rough today, this is clearly the direction AI is moving.
Need a report on your sales pipeline from your CRM system?
Want your email history of interactions with a particular person?
Just ask ChatGPT.
To support all of this, the ChatGPT search interface was changed (yet again). Now, when you chat, you can enable Search Connectors, and then Sources to include in the conversation. Or, when doing Deep Research you can enable Sources to search against.
Choosing Data to Include in Conversations and Search
The Chat Search connectors allow users to search and retrieve information directly from the ChatGPT chat interface. They are suited for quick queries and natural-language search. The user can select which source to search—such as Google Drive or SharePoint—before entering a prompt.
Supported Systems:
Important Detail:
In the case of Google Drive, the connector works by syncing your Drive content into ChatGPT’s index. This makes retrieval faster but also requires IT review of what gets synced and how permissions are enforced. IT teams should test thoroughly before enabling org-wide.
Deep Research uses an AI agent to find the information you need. It allows users to conduct multi-layered queries, scan emails and documents over time, and generate summaries and insights. It’s ideal for working with systems that have higher volume or more complex data.
Supported Systems:
OpenAI is introducing usage credits for Deep Research due to its current monthly limits, with a paid model rolling out soon. Details on how credit tracking and controls will work are still forthcoming.
Whether you lead a small executive team or a 400-person group, my AI Workshops can take your organization from uncertain to AI-advanced—in just one day. Get hands-on, practical training tailored to your needs, and start building a truly AI-enabled organization. Here’s one testimonial:
"Trent’s day-long seminar was one of the most impactful things I've done for my career and company since I can remember. Stellis AI has one of the best training systems for businesses looking to integrate AI safely and effectively."
If the new connections weren’t enough, here are a few additional feature highlights:
ChatGPT’s new Deep Research connector for HubSpot makes it easy to access CRM insights without switching tools. Teams can quickly surface customer histories, deal progress, and support activity—all within ChatGPT—and including features like charting and visualizations on the data.
Sales can prep for meetings faster, support can reference ticket history, and marketing can review engagement—all with simple queries. It will be a time-saver for any team using HubSpot to manage customer relationships.
Allowing ChatGPT to connect via the new Model Context Protocol (MCP) is likely the most significant update in this release. MCP gives organizations the ability to connect ChatGPT to virtually any system—CRMs, databases, wikis, ticketing tools, and more.
This means teams can now pull insights, trigger workflows, or summarize internal data—all through natural language. With MCP, ChatGPT becomes a front-end to your systems, tailored to your business.
The ChatGPT desktop app for macOS now includes the Audio “Record Mode.” It records meetings and generates summaries, key points, and follow-ups. Audio is deleted after processing. Windows and other support is expected soon.
Single Sign-On (SSO) is now supported for ChatGPT Team plans, making it easier to manage user access securely and at scale. This streamlines onboarding and offboarding, aligns with enterprise identity systems, and reduces reliance on individual account setup. Teams can now adopt ChatGPT with centralized control and simplified authentication.
Since it’s confusing which features are available, here’s a table to summarize it. This and other details on connectors are available from OpenAI.
Connector
ChatGPT search
Deep research
Box
✔︎
✔︎
Custom Connectors (MCP)
-
✔︎
Dropbox
✔︎
✔︎
GitHub
-
✔︎
Gmail
-
✔︎
Google Calendar
-
✔︎
Google Drive
-
✔︎
Google Drive synced connector
✔︎
-
Hubspot
-
✔︎
Linear
-
✔︎
Microsoft OneDrive
✔︎
✔︎
Microsoft Outlook Calendar
-
✔︎
Microsoft Outlook Email
-
✔︎
Microsoft SharePoint
✔︎
✔︎
Microsoft Teams
-
✔︎
Connector
Chat search
Deep research
Box*
-
✔︎
Custom Connectors (MCP)*⁺
-
✔︎
Dropbox*
-
✔︎
GitHub
-
✔︎
Gmail*
-
✔︎
Google Calendar*
-
✔︎
Google Drive*
-
✔︎
Google Drive synced connector*
-
-
Hubspot*
-
✔︎
Linear*
-
✔︎
Microsoft OneDrive*
-
✔︎
Microsoft Outlook Calendar*
-
✔︎
Microsoft Outlook Email*
-
✔︎
Microsoft SharePoint*
-
✔︎
Microsoft Teams*
-
✔︎
About Trent: Trent Gillespie is an AI Keynote Speaker, CEO of Stellis AI, former Amazon leader, and advisor on building AI-Native, AI-Enabled businesses. Book Trent to speak to your group or book a call to discuss using AI within your business.
Login or Subscribe to participate in polls.
Did someone forward this newsletter to you? If you're not already signed up, you can subscribe to AI SPRINT™ for free here.