OpenAI has opened up the plug-in system behind ChatGPT to developers, letting them build app-like experiences directly inside the chat interface itself. The move comes as OpenAI pushes deeper into productivity software, with the company’s DevDay presentation on Tuesday laying out new ways for external tools to live alongside ChatGPT’s own Pages, Slides, and Space features.
The core change is simple: developers can now drop their apps into a dedicated sidebar within ChatGPT, complete with interactive panels that let users work with those tools while chatting. OpenAI already connects ChatGPT to everyday tools like Slack, SharePoint, Airtable, and Google Drive, but the new plug-in extensions give each app a permanent home rather than a fleeting link. File viewers are part of the package, supporting whatever formats the developer’s product uses.
The Sidebar Home
The sidebar is the new home for these plug-ins. Developers build their extensions using OpenAI’s new Plugin Creator tool, then submit them to the plug-in directory through a redesigned flow that offers clearer feedback. That streamlined process is meant to make it faster for builders to get their work published and for users to find what they need.
When users open ChatGPT, they will see a list of available plug-ins in the sidebar. Users can click into these panels to interact with the underlying service, whether that means opening a document, reviewing a project board, or managing a workflow.
The file viewers are a notable addition. Because they support whatever formats the developer specifies, plug-in creators can ensure their tools display content natively within ChatGPT rather than forcing users to leave the interface to view files elsewhere. That continuity matters for workflows that rely on documents, images, or other media.
Ranking and Access Controls
OpenAI says it has improved how plug-ins are ranked and recommended within its directory, as well as when they are surfaced in conversations. That means users are more likely to see relevant extensions when they are talking about a topic that matches a plug-in’s capabilities.
The access controls are a practical addition. When a user chooses a plug-in to work with, they can approve its access needs individually. That matters for privacy and security, since some plug-ins may request permission to read files, send messages, or perform other actions on behalf of the user.
The ranking improvements sit alongside the approval flow as part of OpenAI’s push to make the plug-in ecosystem more useful. Better recommendations mean users are less likely to miss a tool that fits their current conversation, while individual approval keeps control over what each plug-in can do.
ChatGPT Sites as Hosts
ChatGPT Sites, the lightweight websites users can build with ChatGPT, can host these plug-ins as well. That opens a second channel for sharing apps, letting someone embed their connected tools into a page that colleagues can use with their own connected data and permissions.
The combination of plug-ins in the sidebar and plug-ins hosted on ChatGPT Sites gives developers two ways to distribute their work. The sidebar is personal, tied to a user’s account. The sites are public, sharable, and independent of any single conversation.
For teams that want to centralize their workflows, the sites offer a natural landing spot. A manager could build a ChatGPT Site that brings together project tracking, document collaboration, and messaging, then share that URL with their team. Everyone who visits would see the same set of tools, each connected to their own data.
The MCP Events Specification
OpenAI is also adding support for the proposed MCP Events specification. That standard would allow plug-ins to start automations based on events in a connected app. The company says it is making automations easier to add and manage, though it did not specify exactly how that improvement works.
OpenAI’s support for the specification positions ChatGPT as a coordination layer rather than just a conversational interface. Developers can build flows that react to real-world changes, turning the chat into a dashboard for ongoing processes rather than a static window into a single app.
The MCP Events support is significant because it extends the plug-in model beyond simple linking. Instead of users manually triggering actions, apps can watch for specific events and respond automatically. A project management tool could notify ChatGPT when a deadline passes or a task is completed, for example, without requiring the user to check the app separately.
The New Plugin Creator Tool
The Plugin Creator is the practical entry point for developers. OpenAI says it is making it easier to build plug-ins, which suggests a simpler setup process and clearer documentation. The redesigned submission flow with clearer feedback should help developers understand where their work stands in the approval pipeline.
The combination of the creator tool and the submission flow points to OpenAI wanting to reduce friction at both ends of the development cycle. Developers get a clearer path to building, and once built, they get clearer signals about whether their plug-in has been accepted.
The new tools are part of a broader pattern of OpenAI lowering barriers for external builders. The company has spent the past year expanding the ways third-party software can integrate with ChatGPT, and the plug-in extensions continue that trend by giving developers a more prominent place to present their work.
What This Means for Users
For individual users, the plug-in extensions mean more context-aware tools inside ChatGPT. Instead of switching between apps, users can keep their workflow in one window, with ChatGPT serving as the hub.
The approval controls are a quality-of-life improvement. Users can decide which plug-ins get access to their data on a per-tool basis, rather than granting blanket permission to everything they install. That granularity helps protect sensitive information from being exposed to tools that don’t need it.
For teams, the combination of sidebar plug-ins and hosted sites offers a flexible playbook. Teams can use the sidebar for day-to-day tasks and the sites for shared projects, mixing the personal and the public as needed.
What This Means for Builders
Developers now have a more structured way to bring their apps into ChatGPT. The Plugin Creator and the submission flow provide a clearer path from idea to deployment, and the improved ranking and recommendation systems should help their work get noticed.
The MCP Events support adds a technical capability that wasn’t previously available. Automations are a feature that many productivity tools offer, and OpenAI’s embrace of the MCP specification means developers can build flows that tie directly into ChatGPT’s conversation model.
The new plug-in extensions are a natural evolution of OpenAI’s existing plug-in system. By giving developers a more structured way to present their work, OpenAI is making ChatGPT a more attractive platform for external software.
Key Facts Box
- DevDay: OpenAI’s presentation on Tuesday announced the new plug-in extensions
- Plug-ins supported: Already connects to Slack, SharePoint, Airtable, Google Drive
- New additions: App-like interfaces, interactive panels, file viewers, MCP Events support
- New tool: Plugin Creator for building plug-ins
- Access controls: Individual approval for each plug-in’s access needs
Schedule Table
| Feature | Timing |
|---|---|
| DevDay announcement | Tuesday presentation |
| Plug-in extensions rollout | Now available to developers |
| Plugin Creator availability | Now available to developers |
| MCP Events support | Now available to developers |
| Improved ranking and recommendations | Now available to developers |
OpenAI’s plug-in system has always been about integration, but the new extensions give developers a dedicated space for their work. The sidebar, the Plugin Creator, and the MCP Events support all point in the same direction: a future where ChatGPT is not just a chatbot, but a platform that hosts a wide range of productivity tools.
The approval controls are a practical guardrail. Users can grant access on a per-plug-in basis, which means they are not exposing their data to everything they install. That balance between openness and control is a sensible design choice, and it should build trust with users who might otherwise hesitate to try new tools.
The plug-in extensions are a clear signal of OpenAI’s ambitions for ChatGPT. The company is positioning the platform as a destination for productivity, not just a conversational assistant. By giving developers a more prominent home for their work, OpenAI is betting that the best applications will win regardless of where they live.
The new plug-in extensions are a step forward for OpenAI’s platform strategy. They give developers a more structured way to present their work, and they give users a more organized way to find and use it. Whether that translates into widespread adoption remains to be seen, but the foundation is now in place.
Source material: “OpenAI expands ChatGPT’s plug-ins with app-like interfaces and automations,” TechCrunch.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.