OpenAI opens a new publishing pipeline for apps running in ChatGPT and pairs it with a ChatGPT app directory so users can actually find what is being built.
For everyday users, the value lies in convenience. Instead of looking for integrations elsewhere, you can browse recommended apps or search for an app’s name and then use it in the same chat you’re already working in.
For developers, this is also a call for direction. If you’ve been thinking about connectors, OpenAI draws attention to apps as the primary way to expand ChatGPT’s capabilities.
Submissions and review begin now
Apps are submitted for review and publication via the OpenAI Developer Platform. The submission flow covers the practical storefront content, including directory listing details, testing guides, and country availability settings. If your app is based on MCP, the process also includes MCP connectivity information.
OpenAI tries to keep quality consistent across all directories. It delivers an Apps SDK in beta, as well as guides on what makes a strong ChatGPT app, sample apps, an open source UI library, and a quickstart. The expectation is for “chat-native” tools that do a job well, not a messy bundle of features.
Discovery switches to chat
The directory is designed to live where the work happens. You can access it in ChatGPT from the Tools menu and it is also available on the web at chatgpt.com/apps.
Once you’ve connected an app, you can run it again by @mentioning it by name or selecting it from the Tools menu. OpenAI is also testing ways to surface relevant apps within a conversation using signals such as context, app usage patterns and user preferences, as well as feedback controls.
Monetization is limited for now
At launch, OpenAI directs developers to link to their own websites or native apps to complete transactions for physical goods. It also says additional monetization options will be explored over time, including digital goods. If apps offer value for you, check out what the best AI chatbots can offer.
The practical takeaway: Users should start with an app that replaces a real routine and see if it saves steps. Developers should ship something tight, submit early, and plan to iterate once directory rollout begins and actual usage shows what sticks.
