ChatGPT has 800 million weekly active users. When someone asks a question, ChatGPT now recommends a single app to help them. Not a list of ten options. One app. Is your brand the one it suggests?

At OpenAI DevDay in October 2025, Sam Altman announced the ChatGPT Apps SDK. This is a new framework that allows developers to build interactive applications that run directly inside ChatGPT conversations. Unlike the old GPT Store where users had to seek out custom GPTs, these new apps get surfaced automatically when a user asks a relevant question. Think of it like organic search, but instead of ranking on a results page, you become the only recommendation.

This post covers everything you need to know about the Apps SDK. You will learn how apps differ from custom GPTs, how automatic recommendations work, what this means for brands, and how to optimize for this new distribution channel.

ChatGPT Active User Figures

In 2023, ChatGPT had 100 million weekly active users. In October 2025, ChatGPT crossed 800 million weekly active users. Understanding how ChatGPT search works is essential for AI visibility.

That is 800% growth in two years.

Remember that report from Siege Media that predicted ChatGPT will overtake Google Search traffic by the end of 2027? It is starting to look less like a prediction and more like a forecast.

The user growth is there. The momentum is there. And now, OpenAI has built the ecosystem to turn ChatGPT into a platform.

The First New App Store In Over A Decade

The Apps SDK is not a plugin system. It is not an integration layer. It is a completely new distribution channel for software products.

The Apps SDK allows developers to embed and monetize their products directly within ChatGPT. It is built on three core pillars.

Connect your data
Trigger actions
Create interactive UI inside ChatGPT

Think about what the App Store did for mobile in 2008. Think about what the Chrome Web Store did for browser extensions. ChatGPT Apps represents that kind of platform shift.

Apps vs Custom GPTs

If you have used the GPT Store before, forget everything you know. ChatGPT Apps are fundamentally different from custom GPTs.

Custom GPTs were essentially personalized chatbots. You configured them with instructions, uploaded some documents, and maybe connected an API endpoint. But they were text-only experiences that lived in a separate tab. Users had to manually visit the GPT Store and pick a bot to chat with.

ChatGPT Apps are interactive applications with real user interfaces. They render inside a sandboxed iframe within the main ChatGPT conversation window. They can display playable videos, form inputs, image galleries, and interactive widgets. They use OAuth for user authentication and make real-time API calls as the conversation flows.

The key difference is discovery. Custom GPTs required users to seek them out. ChatGPT Apps get surfaced automatically within normal ChatGPT conversations.

OpenAI is essentially replacing the GPT Store with the Apps platform. Apps will now focus on bringing real utility by embedding real UI and functionality. Not just wrapped prompts.

Automatic App Recommendations

OpenAI showed two examples at the DevDay keynote that demonstrate how app discovery works.

The first example was a user who uploaded a photo of a hand-drawn sketch. The prompt was "Figma, turn this into a wireframe." The Figma app rendered a fully editable wireframe right there in the chat.

This shows that users can @ mention any app by name. The model understands the context and activates the right tool. No more navigating to a separate store or selecting from a dropdown list.

The second example is the one that matters for SEO.

The prompt was "Make me a playlist for my party based on my favorite songs."

Notice what is missing. The user never mentioned a brand. They never asked for Spotify. They just stated their intent.

ChatGPT automatically recommended an app. It came back with "Use Spotify for this answer."

This is what makes ChatGPT Apps different from every other app marketplace. The platform does not wait for users to browse a directory. It proactively surfaces the most relevant app based on what the user is trying to accomplish.

What This Means For Brands

This is a new type of AI search optimization. The question is no longer "How do I rank number one on Google?"

The question is now "How do I become the default app for an entire category of human intent?"

Winner takes all

This is not a search results page with 10 blue links and a few "People Also Ask" boxes.

It is a single recommendation. One app gets surfaced. The rest do not exist.

A user prompts "Generate a 30-day weight-loss workout routine." Will your fitness app be the one ChatGPT recommends? If it is, that user is yours. The conversion rates will be significant because there is zero competition at the moment of recommendation.

This is a winner-take-all dynamic. Being second place means being invisible.

Embedded customer journeys

When a user interacts with your app, they never leave the ChatGPT interface. They stay in the conversation. They can ask follow-up questions. They can refine their request. They can complete actions.

This means your app can embed your entire customer journey inside ChatGPT. A user can discover your product, understand its value, and convert. All without navigating to your website.

The friction reduction is massive. And reduced friction means higher conversion rates.

Automatic brand visibility

If your app gets recommended for relevant prompts, your brand name and logo appear directly in the ChatGPT interface. Users see your brand at exactly the moment they have a problem you can solve.

This is not like a display ad that users scroll past. This is a contextual recommendation from an AI assistant that millions of people trust. The endorsement value is substantial.

Who Can Build a ChatGPT App?

This opportunity is not limited to big tech companies. Any brand that can create a useful, interactive experience should be building for ChatGPT.

A kitchen company could build an app that lets users visualize new worktops from a simple description. A dog food brand could build an app that gives personalized dietary recommendations based on breed and age. A financial services company could build an app that helps users compare mortgage options in real time.

If you can imagine a tool that helps your customer accomplish a task, you should be building it for ChatGPT.

The apps that win will be the ones that solve real problems. Not the ones that wrap existing content in a chat interface. ChatGPT Apps need to do something. They need to provide utility that justifies the user allowing them into the conversation.

Design Guidelines

OpenAI has published detailed design guidelines for app developers. Understanding these guidelines is important because apps that do not follow them will likely be rejected or deprioritized.

Native design requirements

Your app should blend into the ChatGPT interface. It should not feel like a foreign embed or an advertisement. ChatGPT controls certain global UI elements like the overall background and fonts. Your app needs to work within those constraints.

OpenAI provides a Figma component library with ChatGPT UI components and a style guide. You can create custom components, but they should match the visual language of ChatGPT.

Display modes

ChatGPT Apps use three display modes depending on the complexity of the interaction.

Inline cards are small widgets that appear in the flow of chat. This is the first thing an app shows. They should be concise and focus on a single purpose or a summary. A card might show a result summary with one or two action buttons at most.

Inline carousels let you show multiple items like search results or a list. OpenAI recommends keeping these to 3-8 items. Each item should have a relevant image and title.

Fullscreen views are for more complex interactions or multi-step tasks. Even in fullscreen mode, ChatGPT overlays its chat composer so the user can still converse during the experience. You should use fullscreen sparingly. Only when an inline card cannot accomplish the task.

Upsell restrictions

Your app should feel like a useful tool. Not an advertisement. You cannot show an "Upgrade" button in your header at all times. Upsell prompts should only appear when the user explicitly triggers an action that requires upgrading.

This is an important shift in mindset. The goal is to provide value first. Monetization comes through demonstrating utility. Not through constant prompts.

Optimization Strategies

Your old SEO playbook does not apply here. Here is what you should focus on instead.

Think intents, not keywords

Forget search volume. Focus on the job-to-be-done. What problem is the user trying to solve right now? What task are they trying to accomplish?

Your app needs to be the solution to a specific category of intent. The more clearly your app solves a defined problem, the more likely ChatGPT will recommend it for relevant prompts.

Your app is your new content

A 2,000-word blog post will not get you recommended. An interactive tool that solves the user's problem live in the chat will.

Think of your app as your new pillar page. It is the asset that demonstrates your expertise and provides value. The difference is that it does not just explain how to solve a problem. It actually solves the problem.

Optimize your app listing

A new form of App Store Optimization is emerging. The ranking factors are not public, but we can make educated guesses about what will matter.

Clarity. Does your app name and description match the intents it serves? If your app helps with video editing, does the listing make that obvious?
Functionality. Does your app deliver a good user experience? Apps that frustrate users will likely get deprioritized.
Engagement signals. Usage volume, session length, and user ratings will likely influence recommendations. Apps that users love will get surfaced more often.

Submission Timeline

The ChatGPT Apps platform is currently in preview. Only a limited set of partner apps are publicly available. Spotify, Booking.com, Expedia, and a few others were part of the early pilot.

OpenAI has communicated that they plan to open app submissions to all developers by the end of 2025. The exact date is not public yet.

Once submissions open, approved apps will get listed in a dedicated App Directory inside ChatGPT. Users will be able to browse and search available apps. Think of it like the App Store, but inside ChatGPT.

OpenAI has also announced plans for monetization through something called the Agentic Commerce Protocol. This will allow users to complete purchases and subscriptions directly inside ChatGPT. The details are not fully public yet, but this signals that ChatGPT Apps will become a revenue channel. Not just an acquisition channel.

If you want to be among the first apps in the store when it opens, you should be building now.

Tracking App Rankings

The ranking factors for ChatGPT app recommendations are not yet understood. But we are building the tools to figure them out.

At LLMrefs, we are tracking which apps get surfaced for which prompts directly inside ChatGPT. We are building the analytics platform for this new era of AI search.

If you want to understand how your brand appears in AI search and claim your spot in this new marketplace, get in touch.

ChatGPT Apps SDK for SEOs and developers