Introduction

ChatGPT can now directly interact with your favorite services. Want Spotify to create a party playlist? Need Zillow to pull up housing listings? Or perhaps you’d like Canva to turn your outline into a slide deck? All of this can be done without leaving the chat. OpenAI is revamping its ChatGPT app strategy, integrating real applications into ChatGPT and providing developers with fresh tools to build and monetize these applications. If you remember the initial wave of custom GPTs and the brief existence of the GPT Store, consider this a sequel with a clearer focus and improved infrastructure.

In this guide, we dissect what has changed, what you can do now, how privacy is handled, and why this initiative is significant for developers, businesses, and anyone intrigued by the forthcoming interface changes.

What Exactly Did OpenAI Change?

In brief: Apps now reside within ChatGPT and function as built-in assistants, ready to be summoned by name. Additionally, ChatGPT can recommend an app when it identifies that one could be beneficial. The initial lineup includes Spotify, Zillow, Canva, Booking.com, Coursera, Expedia, and Figma, with more partners on the horizon. Developers have access to a new Apps SDK, a review process is in the works, and monetization details are expected later this year.

OpenAI’s updated app model leverages the Model Context Protocol, an open standard designed for connecting AI to external tools and data, enhancing interoperability compared to previous “plugins” and simplifying integration across different platforms.

Why This Reboot Matters

• Enhanced Discovery and Usability: In the prior GPT Store iteration, while power users experimented, casual users often struggled to find the right GPT when needed. Now, apps can show up naturally in conversations or be called upon by name, thereby lowering the barriers for non-experts.
• Alignment with Open Standards: By building on the Model Context Protocol, developers can use the same connectors across various platforms, avoiding vendor lock-in. This could accelerate adoption and cultivate a healthier app ecosystem.
• Monetization Opportunities: A structured submission and review process is being developed, with OpenAI promising to share revenue details soon. Coupled with Instant Checkout and the Agentic Commerce Protocol, this strategy sets the stage for in-chat transactions.
• Access to a Massive User Base: OpenAI is targeting over 800 million ChatGPT users, establishing one of the largest potential distribution channels for interactive apps.

How Apps Work Inside ChatGPT

Think of ChatGPT as your personal conductor. You can call an app by typing its name, like, “Spotify, create a 90s throwback playlist for a two-hour road trip.” The first time you connect an app, ChatGPT will prompt you for permission regarding the data to share. After that, the conversation can seamlessly flow between ChatGPT and the app, featuring interactive UI elements such as maps, music playlists, or presentation slides embedded in the chat.

OpenAI showcased this functionality in real-time demos, featuring Canva generating a poster and a pitch deck, alongside Zillow providing an interactive map of homes for sale that users could refine with further questions. Expect this experience to expand as more applications are introduced.

What’s Available Now and What’s Coming Next

Available Now

• Apps from: Spotify, Zillow, Booking.com, Canva, Coursera, Expedia, and Figma, available in regions where these services operate, initially in English.
• Audience: All logged-in ChatGPT users outside the EU, across Free, Go, Plus, and Pro plans.

Coming Soon

• Additional apps from partners such as DoorDash, OpenTable, Target, Uber, and others, alongside a public submission process for developers.
• A browsable app directory, quality tiers for well-designed apps, and monetization options.
• Rollout to ChatGPT Business, Enterprise, and educational accounts, plus European availability as OpenAI finalizes necessary work.

Privacy, Security, and User Control

OpenAI has stated that each app must adhere to its usage policies and be suitable for all audiences. Upon the initial connection of an app, ChatGPT will provide a consent flow so you can specify what information to share. Developers are required to maintain clear privacy policies and limit data collection to what’s necessary. More granular permission controls are being developed, and OpenAI is enhancing administrative tools as apps roll out for Business, Enterprise, and educational use.

How This Differs from Plugins and the GPT Store

If you experimented with ChatGPT’s “plugins” in 2023 or browsed the GPT Store in early 2024, you encountered an initial version of this concept. Plugins felt experimental and cumbersome; GPTs were simple to create but difficult to discover, and the business model remained unclear. Initial excitement has waned, leading many GPTs to become internal tools rather than consumer-friendly experiences. The new reboot replaces this with a more straightforward invocation model, a unified protocol, a formal SDK, and an upcoming directory and review process.

The Standards Layer: Why MCP Matters

The Model Context Protocol is foundational. Instead of having ChatGPT and each app communicating in separate dialects, MCP provides a common language for connecting AI agents to tools, files, browsers, and services. This reduces complexity for developers who want their applications to function across various AI platforms, not just ChatGPT. Major AI companies and the open-source community have embraced MCP since spring, and the ChatGPT Apps SDK builds directly atop it.

A New Path to In-Chat Commerce

Apps become exponentially more functional when they can facilitate transactions, including purchases. In late September, OpenAI rolled out Instant Checkout, powered by the Agentic Commerce Protocol and built with Stripe. Users in the U.S. can already buy items from Etsy sellers directly within ChatGPT, with support for additional merchants on the way. Importantly, product results are organic and not sponsored, and merchants incur a small fee on completed sales. This commerce layer is expected to integrate into ChatGPT apps once OpenAI opens submissions and monetization later this year.

For Developers: What the Apps SDK Includes

OpenAI’s Apps SDK is currently available for preview. It extends MCP, allowing developers to define both the logic and interface of their app, connect it to ChatGPT users, and link directly to their backend. The model manages natural language while your app presents interactive features in the chat. Documentation, example applications, and development guidelines are available, and OpenAI plans to begin reviewing and publishing apps later this year. Highlights include:
– Build once on an open standard; reuse MCP servers across ChatGPT and other compatible tools.
– Scale your distribution, tapping into a potential audience of over 800 million ChatGPT users.
– Prepare for monetization, with OpenAI anticipated to share further details soon, including support for its agentic commerce.

Developers and teams focused on creating agents should also note two related updates from DevDay:
– Codex is now generally available, featuring a Slack integration, an SDK, and admin tools to assist engineering teams in automating coding tasks using OpenAI’s coding agent.
– AgentKit is a toolkit for designing, embedding, and evaluating agent workflows, including a visual Agent Builder, ChatKit for agent UIs, and expanded evaluation and reinforcement fine-tuning.

What This Means for Apple and Google

OpenAI’s latest move introduces an app directory, developer submissions, and in-chat purchases to a platform catering to hundreds of millions of users. Although it doesn’t function as a traditional mobile app store, the combination of app discovery, distribution, and commerce within ChatGPT is beginning to mirror a marketplace that could compete for developers’ attention and users’ time. Keep an eye on how Apple and Google might respond, particularly regarding native integrations and policies for apps embedding ChatGPT experiences.

Real-World Examples to Try

• Travel Planning: Use Booking.com or Expedia within ChatGPT. Request, “3 nights in Paris, two adults, Nov 21-24, near the 6th arrondissement,” and refine by price or amenities.
• Design Creation: Collaborate with Canva. Provide an outline and instruct Canva to convert it into slides. Follow up with adjustments like, “make slide 3 more visual.”
• Home Searching: Check out Zillow. Ask for “3-bedroom homes under $600k in Pittsburgh near good schools” and filter options in chat.
• Playlist Generation: Request Spotify for a workout mix. Ask, “Spotify, create a 45-minute HIIT playlist at 160-175 BPM.” Connect once, and it’s reusable anytime.

What to Watch Next

• EU Rollout Timing: Apps are now live for users outside the EU, with European availability anticipated once OpenAI completes necessary preparations.
• App Review and Quality Standards: OpenAI is developing a directory and may showcase the best-designed apps more prominently, raising quality expectations.
• Safety and Permissions: Look for more precise data controls, clearer privacy policies, and improved governance as apps become available for Business, Enterprise, and Edu.
• Monetization Mechanics: The specifics of revenue sharing, pricing models, and the role of the Agentic Commerce Protocol will shape whether building a ChatGPT app becomes viable.
• Interoperability Beyond ChatGPT: As more vendors adopt MCP, developers will need to consider how to maintain cross-platform AI apps without substantial duplication of work.

How We Got Here: A Quick Timeline

• Nov 2023: OpenAI introduces custom GPTs and hints at a store during its inaugural DevDay. Initial excitement is evident, but details around revenue and curation remain vague.
• Apr 2025: The Model Context Protocol gains traction as a straightforward method for connecting AI with everyday apps and data, receiving broad ecosystem support.
• Sep 2025: OpenAI launches Instant Checkout and an open commerce spec with Stripe, laying a foundation for in-chat purchases.
• Oct 2025: OpenAI brings real apps into ChatGPT, along with an Apps SDK and plans for submissions, a directory, and monetization.

Bottom Line

This reboot represents OpenAI’s most coherent effort yet to transform ChatGPT into a platform where apps seamlessly integrate into conversations. The experience is more intuitive for users, standardized for developers, and finely tuned for discovery and commerce. If OpenAI successfully addresses review quality, safety, and monetization, ChatGPT could evolve into a daily resource for not just answering questions but also completing tasks.

FAQs

What is the difference between ChatGPT apps, plugins, and custom GPTs?

Plugins were early integrations that felt attached rather than cohesive. Custom GPTs allowed individuals to create specialized chatbots, but they faced challenges in discovery and monetization. In contrast, ChatGPT apps are developed using an SDK on a shared protocol, featuring interactive chat interfaces and residing in a directory with established review standards.

Are ChatGPT apps available everywhere?

Currently, apps are available to logged-in users outside the EU, with European availability anticipated later. Partner apps are functional in regions where those services operate, primarily starting in English.

How will developers make money with ChatGPT apps?

OpenAI plans to reveal monetization details later this year, while also constructing agentic commerce pathways for instant checkout within ChatGPT. A structured review process and directory will elevate visibility for the best applications.

What data do apps access when I connect them?

Upon the initial use of an app, ChatGPT will prompt you to connect and clarify what data may be shared. OpenAI obliges developers to have clear privacy policies, minimize data collection, and ensure transparency. More detailed controls are planned for the future.

What types of apps can I use right now?

Apps available include Spotify for music playlists, Zillow for real estate listings, Canva for design, Booking.com and Expedia for travel, Coursera for learning, and Figma for design collaboration. More will be added over time.