By skyforbes 
Remember when ChatGPT was just a chatbot? Yeah, those days are officially over.
At DevDay 2025 in San Francisco, OpenAI CEO Sam Altman took the stage and dropped what might be the biggest shift in ChatGPT’s identity since its launch: Apps inside ChatGPT. And I’m not talking about simple integrations or plugins. I’m talking about full-fledged, interactive applications that live and breathe inside your chat window.
Let’s unpack what just happened, because this is huge.
The Numbers That Made Everyone’s Jaw Drop
Before diving into the apps themselves, Altman shared some mind-blowing growth stats. Back in 2023 at the first DevDay, ChatGPT had about 100 million weekly users and 2 million developers building with OpenAI’s tools.
Fast forward to today? Over 800 million weekly users and 4 million developers. Oh, and their API throughput? It jumped from around 300 million tokens per minute to over 6 billion.
Yeah. Billion with a B.
These aren’t just vanity metrics — they show that OpenAI has built something developers actually want to build on top of, and users actually want to use. And now, they’re opening the floodgates even wider.
So What Are “Apps Inside ChatGPT” Exactly?
Here’s the vision: OpenAI wants ChatGPT to be more than just a conversational AI. They want it to be a powerful tool for productivity, invention, and learning where you can get stuff done, not just have conversations.
The new Apps SDK (which launched in preview at DevDay) lets developers build applications that are:
- Interactive — Not just static responses, but fully functional UIs
- Adaptive — They respond to context and user needs in real-time
- Personalized — They can connect to your data and preferences
Think of it like this: instead of ChatGPT just telling you about something, it can now show you and let you interact with it, all without ever leaving the chat.
The Tech Behind It: Apps SDK
OpenAI built the Apps SDK as a complete full-stack solution for developers. Here’s what’s in the toolkit:
Connect Data — Apps can tap into your existing accounts and data sources (with your permission, of course)
Trigger Actions — They can actually do things, not just display information
Render Interactive UI — Full, beautiful interfaces right in the chat window
The SDK is built on MCP (Model Context Protocol), which is an open standard. This means developers have full control over both their backend logic and frontend UI design. No compromises.
And here’s the kicker: apps built with this SDK can instantly reach hundreds of millions of ChatGPT users. For developers, that’s the kind of distribution that would normally take years to build.
Finding and Using Apps: Two Ways
OpenAI designed app discovery to feel natural, not forced. There are two main ways you’ll encounter apps:
1. Call Them by Name
You can explicitly invoke an app just by mentioning it. Want to turn a sketch into a proper diagram? Just say “Figma, turn this sketch into a workable diagram,” and boom — Figma appears in your chat.
2. Smart Recommendations
This is where it gets interesting. ChatGPT can contextually suggest apps based on your conversation. Planning a party and asking for music suggestions? ChatGPT might suggest opening Spotify to create that playlist right then and there.
It’s like having an AI assistant that not only understands what you need but also knows which tool would work best for the job.
The Demos That Stole the Show
Okay, descriptions are cool and all, but the live demos are where this really clicked. OpenAI showed off three partner apps, and each one demonstrated something different about what’s possible.
Coursera: Learning Gets Interactive
The demo started simple: someone asked the Coursera app in ChatGPT to help them learn about machine learning.
First-time users have to give consent to connect the app (privacy first, always), and then the magic begins. The app displays inline by default — meaning it shows up right in your chat thread, not in a separate window.
Here’s where it gets cool: video playback inside the chat. Not a link to a video. Not “open this in a new tab.” The video plays right there, with picture-in-picture and fullscreen modes.
But wait, there’s more. When you’re watching a video, it pins to the top of your screen so you can keep scrolling through the conversation. Want to ask a question about something mentioned in minute 3:45 of the video? Go ahead. The Apps SDK provides an API that lets the app pass precise interaction context back to ChatGPT.
So when you ask “Can you explain that gradient descent concept mentioned in the video?” ChatGPT knows exactly what you’re talking about because the Coursera app told it where you are in the video. This is what they mean by “talking to apps” — it’s a two-way conversation.
Canva: From Chat to Creative Assets
Next up was Canva, and this demo was all about creation.
The presenter asked Canva to create a poster, and Canva pulled context from the ongoing conversation to understand what kind of poster was needed. Within seconds, a fully-designed poster appeared in the chat.
You can view it inline or go fullscreen to see it in detail and make tweaks. The interface isn’t some watered-down mobile version — it’s the real Canva experience, embedded in ChatGPT.
Then came the plot twist: “Now turn this poster into a seed-round pitch deck.”
And Canva just… did it. Right there in ChatGPT. The poster transformed into a multi-slide pitch deck, ready to present to investors. No exports, no switching between tools, no copying and pasting. Just pure, seamless creation.
This is the kind of workflow that used to require multiple apps, multiple logins, and way too many browser tabs. Now it’s one conversation.
Zillow: Real Estate Meets Real Intelligence
The Zillow demo showed how apps can compose context with other ChatGPT tools to create truly intelligent experiences.
It started with ChatGPT recommending cities for business expansion based on conversation history. Once a city was chosen, the Zillow app was invoked to show homes for sale there.
An interactive map appeared right in the chat. Not a screenshot. A live, clickable, scrollable map with property listings.
You can view it fullscreen to get the full Zillow experience, filter results (like “show me three-bedroom homes with a yard”), and browse listings as if you were on Zillow’s website.
But here’s the genius part: the Zillow app exposes its context to ChatGPT. So when you ask a detailed question like “How close is this house to a dog park?” ChatGPT can compose information from Zillow (the house you’re looking at) with other tools like search to give you a comprehensive answer.
This is contextual intelligence at work. The app isn’t isolated — it’s part of the broader ChatGPT ecosystem, working together with other tools to help you make decisions.
Show Me the Money: Monetization Plans
OpenAI isn’t leaving developers hanging on the business model question. They’ve outlined several monetization paths:
Existing Subscribers — If someone already pays for your service, they can sign in from ChatGPT and access their account seamlessly
Agentic Commerce Protocol — This is the fancy name for instant checkout inside ChatGPT. In the future, users will be able to purchase things without ever leaving the conversation
Featured Placement — Apps that meet higher quality standards for design and functionality will be featured more prominently in the app directory
The message is clear: build great apps, and OpenAI will help you reach users and make money.
The Developer Journey: What’s Available Now
Here’s where things stand for developers who want to start building:
Available Today:
- Apps SDK in preview (start building now)
- Select partner apps already live in ChatGPT
- Developer guidelines published in draft form
Coming Later This Year:
- App submission process opens up
- Public app directory launches
- Apps will be browsable and discoverable for all users
The phased rollout makes sense — OpenAI wants to gather feedback from early builders and refine the experience before opening the floodgates completely.
Why This Matters More Than You Think
Let’s zoom out for a second and think about what this really means.
ChatGPT started as a text box where you ask questions and get answers. Then it got browsing, image generation, and data analysis. But this? This transforms ChatGPT into a platform.
A platform where developers can build businesses. Where users can get work done without juggling a dozen different apps. Where the conversation itself becomes the interface for everything you need to do online.
For developers, it’s unprecedented distribution. Imagine launching a product and immediately having access to hundreds of millions of potential users. That’s the pitch.
For users, it’s a fundamentally simpler way to interact with technology. Instead of remembering which app does what, you just… talk. And the right tools appear when you need them.
The Bigger Picture: OpenAI’s Platform Play
This isn’t just about apps. It’s about positioning.
With this move, OpenAI is going head-to-head with app stores, browsers, and operating systems. They’re saying “What if the interface for everything was a conversation?”
Google has Search. Apple has the App Store. Meta has social networks. OpenAI wants ChatGPT to be the layer where you interact with everything.
It’s ambitious. It’s bold. And based on the demos we saw at DevDay, it might actually work.
What’s Next?
The Apps SDK is in preview now, which means developers can start experimenting immediately. If you’re a developer, now is the time to start thinking about what kind of app you could build and how it might fit into a conversational interface.
If you’re a user, keep an eye on your ChatGPT sidebar. Over the coming months, you’ll start seeing apps appear that can do way more than you ever thought possible in a chat window.
One thing’s for sure: ChatGPT isn’t just a chatbot anymore. It’s becoming the platform for the AI era.
And we’re all invited to build on it.
What apps would you want to see inside ChatGPT? Drop your thoughts in the comments — I’d love to hear what you’d build with this kind of reach.
Follow me on @iamtanujsharma