The future of ChatGPT for businesses is here, and this new update – connectors that connect to your company knowledge – could pave the way for simpler, more efficient ways of working with some of the most used and notable work tools.
This update, introduced for ChatGPT Team and Enterprise, enables ChatGPT to connect directly to tools such as databases, CRM systems, and communication platforms.
What Are ChatGPT Connectors?
OpenAI announced yesterday that it would be introducing new connectors that allow organizations to connect their work context directly to ChatGPT, its primary AI offering. This is enabled through the ability for organizations to connect their internal knowledge sources to ChatGPT, allowing anyone in the organization to find the “most relevant information at the right time”, according to OpenAI.
ChatGPT can now pull in the appropriate context from a range of tools, including:
- SharePoint
- Outlook (Mail and Calendar)
- Teams
- Google Drive
- Gmail
- Google Calendar
- Dropbox
- Box
- HubSpot
- GitHub
- Linear
These connectors are currently in beta and are available in ChatGPT and deep research for Team, Enterprise, and Edu plans. Pro and Plus plans have access to these connectors only in deep research.
Are There Any Security Concerns?
Nate Gonzalez, OpenAI’s Product Leader, explained that although these connectors work with an organization’s unique knowledge sources, security remains a top priority.
“We respect all of the existing user permissions inside your organization, and we keep your data private and secure within the team and enterprise workplaces,” he said.
OpenAI’s official documentation also mentions: “When ChatGPT sends a query to connected apps, it may also leverage relevant information from memories to make the query better and more useful.”
Additionally, ChatGPT can utilize relevant information accessed from connectors to inform its search queries when generating responses online. Users can choose to disable this function or control specific memories.
OpenAI also makes it clear that for ChatGPT Team, Enterprise, and Edu customers, the information obtained from the connectors is not used to train its models. For Free, Plus, and Pro users, the company states that it “may use information accessed from connectors to train our models if the ‘Improve the model for everyone’ setting is on.”
Users on these lower-tiered plans should be mindful of their security agreements, taking the time to ensure that their data isn’t used in a way that infringes on their privacy.
Can You Build Your Own Connectors?
OpenAI has also announced that, in addition to the company’s own, pre-built connectors, workspace admins can set up custom deep research connectors to any data source your company cares about using an MCP server.
With this feature, the option to add custom connectors that follow the Model Context Protocol (MCP) becomes available to connect to custom third-party apps and internal sources.
These are intended for developer use only, as OpenAI is not able to verify them.
Summary
This connection framework not only presents a mighty step forward for OpenAI but also a new venture for the entire LLM and generative AI industry.
It is interesting to note that OpenAI does not currently have a connector for Salesforce; however, given their existing partnership and backend collaboration, it would be reasonable to predict that a front-end connection will be forthcoming soon.
However, ultimately, if OpenAI isn’t the company to do this first, another company probably will. Reducing the complexity of software and work tools will be on the agenda for plenty of companies, and it is something that admins and developers alike should be prepared for.
It will be exciting to see which new platforms OpenAI will build connectors for next, as well as which neighboring AI platforms follow suit.
