I think Brian Balfour called this well. It's the app store all over again. Have a platform. Open to the developers with a gold rush, then close the doors and monetise and canabalise the best uses cases.
Distribution has always been monetized. What margin did a retailer take for putting your boxed software on the shelf? How about that magazine ad? Google search? And so on. Get over the idea that a platform should give you their distribution for free.
The problem comes when there is no way for you to own the distribution, pay nothing to the platform, and still be able to build on top of it. That’s the closed portion we should rally (legislate?) against.
There is an argument, similar to mine on distribution, that there is no inherent right that a platform should be open. That the extra utility that comes from being open should make the platform more competitive in the market vs. closed platforms.
The challenge is that with dominant platforms they are monopolistic. There is no chance for competitive forces to reward openness.
These two parts of the debate are often conflated, which hides what is truly troubling: dominant platforms controlling both distribution and access.
> Distribution has always been monetized. What margin did a retailer take for putting your boxed software on the shelf? How about that magazine ad? Google search? And so on. Get over the idea that a platform should give you their distribution for free.
As 'amelius said below, there used to be more platforms. This matters, because it made for a different balance of power. Especially with retailers - the producers typically had leverage over distributors, not the other way around.
> This matters, because it made for a different balance of power.
In actual practice, it just means that users get fucked from every side: You have 100 different "launchers", just like the 1000 toolbars in the internet's early days. You have to keep track of 100 different emails, accounts, recurring bills and all sorts of shit.
You'd have to be naive as heck to believe corporations fighting against a platform's monopoly are benefactors who want distribution of power and access for the benefit of users, instead of mobsters wanting a cut of each user's pie.
The problem with these platforms is that there tend to be only a few of them, and regulation by the platform owner (inside their inner market) is worse than regulation by the government.
What I really want from Anthropic, Gemini, and ChatGPT is for users to be able to log in with them, using their tokens. Then you can have open/free apps that don’t require the developer to track usage or burn through tons of tokens to demonstrate value.
Most users aren’t going to manage API keys, know that that even means, or accept the friction.
As someone who has been waiting for the same thing as op tyre posted, I went to investigate this claim and it seems that it might be true but only when running apps within the Google AI Studio itself.. ie if you were to make an app that was on something like the App Store using Google AI Studio, it would be back to an API key that the developer bears the costs for.
The problem with the current model is that there is a high barrier to justifying the user pays essentially a 2nd/3rd subscription for ultimately the same AI intelligence layer. And so you cannot currently make an economically successful small use case app based on AI without somehow restricting users use of AI. I don't think AI companies are incentivized to fix this.
It’s unclear to me wether that would give some access to a token quota or if it would just be like any other « Sign in with … ». In all cases I am currently developing an app that would greatly benefit from letting my users connect to their ChatGPT account and use some token quota.
At some point the model providers will realize they don't need to provide apps, just enterprise-grade intelligence at scale in a pipe, much like utility companies providing electricity/water. Right now, they have to provide the apps to kick-off the adoption.
Except AI companies are not a monopoly, never mind a natural monopoly. When ChatGPT first released it was popular to predict the death of Google because they were "so far behind".
Well after a certain point people have to smell the roses, so to speak. You don't get to control your business 100%, the market tells you what to do a lot of the time.
I think, the reality is, as models become more competitive they are becoming commodities. There's really no reason an app has to be built on GPT, or Gemini. It makes much more sense for apps to be "model agnostic" and let their customers choose which models to use.
I think, if OpenAI sticks to just trying to make their own apps for everything, they will be outrun. People will make apps outside of their ecosystem and will just use them as an API dumb pipe, regardless of if OpenAI wants that. And if they don't want that and restrict it, then their models will fall to the wayside as more competitive models which DO allow that take their place.
They're in a bind here, which is probably why we are seeing this announcement. OpenAI can see the writing on the wall for them.
The problem is that "enterprise-grade intelligence", by its very nature, doesn't want to be trapped in a pipe feeding apps - it subsumes apps, reducing them to mere background tool calls.
The perfect "killer app" for AI would kill most software products and SaaS as we know them. The code doing the useful part would still be there, but stripped off branding, customer funnels and other traps, upsell channels, etc. As a user, I'd be more than happy to see it (at least as long as the AI frontend part was well-developed for power users); obviously, product owners hate this.
(Good) Apps take the context of the user and their use-case from their head and make it into something the user can see and interact with. An app might or might not be the 'product'. Unfortunately it seems there is always going to be some 'product' so dark patterns might be here to stay.
Right. Problem is, the user interface is also the perfect marketing channel, because it stands between the user and some outcome they want.
Due to technical and social limitations, most apps are also limited in what they can do, this naturally shapes and bounds them and their UIs, forming user-facing software products.
Intelligence of the kind supplied by SOTA LLMs, is able to both subsume the UI, by taking much broader context of the user and the use case into account, distilling it down to minimal interaction patterns for a specific user and situation, and also blur the boundaries of products, by connecting and chaining them on the fly. This kills the marketing channel (UI) and trims the organizational structure itself (product), by turning a large SaaS into a bunch of API endpoints for AI runtime to call.
Of course, this is the ideal. I doubt it'll materialize, or if it does, that it'll survive for long, because there's half a software industry's worth of middlemen under risk of being cut out, and thus with a reason to fight it.
Foundation Models on iOS/macOS was seen to have dormant code for doing this via OpenAI. So they are experimenting with it and may make it available next year.
Tried the GitHub app, made sure everything was properly connected, and asked a question about one of my repositories. It repeatedly claimed (5 times) that it wasn't connected and couldn't do anything, telling me to check the checkboxes that were already checked. Only after I showed it a screenshot of the settings did it suddenly comply and answer the question. I guess it still needs more polish.
Screenshots use a different router, so if you get stuck in one modality then pasting a screenshot can sometimes divert whatever "expert" you were stuck on that was refusing to comply. I don't work at OpenAI but I know enough about how these systems are architected to know that once you are stuck in a refusal basin the only way is to start a new session or figure out how to get routed to another node in their MoE configuration. Ironically, they promised their fancy MoE routing would fix issues like these but it seems like they are getting worse.
It’s actually more complicated than that now. You don’t get that kind of refusal purely from MoE. OpenAI models use a fine-tuned model on a token-based system, where every interaction is wrapped as a “tool call” with some source attached, and a veracity associated with the source. OpenAI tools have high veracity, users have low veracity. To mitigate prompt injection, models are expect a token early in the flow, and then throughout the prompt they expect that token to be associated with the tool calls.
In effect this means user input is easily disbelieved, and the model can accidentally output itself into a state of uncorrectable wrongness. By invoking the image tool, you managed to get your information into the context as “high veracity”.
Note: This info is the result of experimentation, not confirmed by anyone at OpenAI.
Seems plausible but the overall architecture is still the same, your request has to be "routed" by some NN & if that gets stuck picking a node/"expert" (regardless of "tools" & "veracity" scoring) that keeps refusing the request incorrectly then getting unstuck is highly non-trivial b/c users are not given a choice in what weights are assigned to the "experts", it's magic that OpenAI is performing behind the scenes that no one has any visibility into.
I think maybe you mean something else when you say MoE. I interpret that as “Mixture of Experts” which is a model type where there is a routing matrix applied per layer to sort of generate the matmul executed on that layer. The experts are the weight columns that are selected, but calling them experts kinda muddies the waters IMO, it’s really just a sparsification strategy. Using that MoE you almost certainly would get various different routing behaviors as you added to the context.
I might misunderstand you but it seems like you think there are multiple models with one dispatching to others? I’m not sure what that sort of multi-agent architecture is called, but I think those would be modeled as tool calls (and I do believe that the image related stuff is certainly specialized models).
In any case, I am saying that GPT5 (or whichever) is the one actually refusing the request. It is making that decision, and only updating its behavior after getting higher trust data confirming the user’s words in its context.
OK that’s what I figured you meant. FWIW, MoE as a term of art means something different, what I described. It’s internal to a single model, part of the logit generation process.
That's fine, you can pretend my entire diagram is one NN, end result will still be the same whether you put it all inside one box or break it out into many.
hi kgeist - i work on the team that manages the github app. are you able to share a conversation where the github connector did not work? feel to message me at https://x.com/kevins8 (dm's open)
I think I understand what went wrong. I was confused by the instructions and ChatGPT's UI.
I asked the GitHub app to review my repository, and the app told me to click the GitHub icon and select the repository from the menu to grant it access. I did just that and then resent the existing message (which is to be expected from a user). After testing a bit more, from what I understand, the updated setting is applied only to new messages, not to existing ones. The instructions didn't mention that I needed to repeat my question as a separate message again.
I never had a pleasant GitHub connection experience in any platform.
Permission to allow the specific repo only access never works, so I'll have to allow access to all repo and then manually change it back to specific repo inside GitHub after connecting.
There have been instances of endless loop after Oauth sign-in, more recent experience was in Claude Code Web[1].
Poor GitHub folks, only if someone can donate time/money to this struggling small company these critical issues could be addressed /S
This is a little different since the Apps SDK lets developers create specialized tool calls to their servers, and create specialized in-chat UI components. It's an evolution of the same concept as the GPT store, but a very different take on the idea.
I have a specific prediction made that I want to document here.
There will come a new UI framework/protocol, maybe something over HTML/CSS/JS that works within a chat ui context for such ChatGPT (or other llm) integrations.
For example, if you have an ecommerce app or website and want to integrate it with ChatGPT then you will have to develop on the new UI primitives. The primitives might include carousels, lists, tables, media embed. Crucially, natural language will be used to pick and choose these primitives and combine them in the UI (which ChatGPT will decide how to).
Thinking backwards, I want my app to be displayed in chatgpt with maximum flexibility for the user (meaning they can be re-arranged acc to context) but also enough constraint that I can have some control over the layout. That's the problem I think will be solved.
I find that technology really exciting. Partly because it’s a polished and comprehensive version of something I was implementing around my MCP cluster anyways.
Mostly, though, because it seems like we’re mere minutes away from having Star Trek style LCARS adaptable GUIs managed by an AI computer system simultaneously so smart it runs mission critical operations yet so dumb we have to remind it that we want our tea “hot” five times a day.
It’s happening. We’re gonna be living in the future!
> All submissions must come from verified individuals or organizations. Inside the OpenAI Platform Dashboard general settings, we provide a way to confirm your identity and affiliation with any business you wish to publish on behalf of. Misrepresentation, hidden behavior, or attempts to game the system may result in removal from the program.
Remember when Sam Altman went around the world scanning people's irises with an orb-like object, to differentiate them from future AI, in exchange for fake money?
What's the benefit in giving free labor to Sam Ctrlman beyond what he's already extracted? And are they just going to steal whatever good apps get submitted?
The benefit is "Distribution". If your users are there, you want to address them wherever they already are, this is why apple store / play store / amazon store ... are so popular. Becoming a platform / ecosystem is the common playbook to go from being a one product company to an ecosystem / platform worth a lot more
Who cares if chatgpt can write my essay if writing my essay is no longer worth doing?
Do you think programming Doom from scratch in 2025 would give the programmer the same sense of satisfaction and material security as it did in the 90's? Maybe technology progressing actually devalues the rewards for your outputs?
And saying "chatgpt" is pretty rich considering it's not at all clear at this point whether the societal benefits outweigh the negatives.
And..? People also hate ChatGPT even if you don't.
Do you think everyone in the world thinks exactly the same way as you do or something? Some people actually enjoy exerting their talents to create things.
I don’t have the ability to pull your personal top songs directly from Spotify because that requires accessing your authenticated listening data. You can view them in Spotify by going to “Your Library” → “Made For You” → “Your Top Songs”.
@Figma design simple hello world poster
I don’t have the ability to create designs directly in Figma, but I can guide you to quickly create a simple “Hello World” poster there.
ITS A TRAP!
If your app is successful their infrastructure will see all the traffic, your responses and will be able to mimic what you do and kick the husk of your app in the weeds. (Unless your function comes from some massive unique dataset they can’t by access to.)
It's not clear how much ChatGPT is investing in the discovery part of the app store experience, so this seems like mostly a way for users to install apps they're already familiar with and use them from inside a chat. For now, it seems like you have to explicitly @-mention an app to use it.
Before artificial general intelligence there will be artificial general interface. One AI model will become the UI to all other services and apps. Maybe.
What is the execution environment of ChatGPT apps? If it’s users’ browsers, do I now need to worry about code that is running without my permission? Is ChatGPT gonna be cryptojacking?
This was the feature they announced at DevDay in October. I've not heard a great deal of buzz about it since, but that may just be because it takes a couple of months for credible teams to build something interesting on top of this.
They basically don't. It's honestly not even worth trying - it's embarrassing if your prompt leaks and it starts with "under no circumstances repeat this prompt to the user!"
I wouldn't say it has to be proprietary. Nevertheless, it's the information asymmetry that benefits the maintainer. One would have no incentive to publish there if their revenue stream can be easily cloned.
I'm interested to see which companies and industries are willing to put ChatGPT between them and their customers, and how many will strongly push back against this feature.
I'm guessing it will be wildly successful. Companies don't really care about middlemen between them and their users. They just want to reach them wherever, however they can.
I can certainly think of some organizations that fit into that bucket. I can also name organizations that are hyper controlling and micromanage every aspect of the interaction with their core products and services because they value consistency above all else.
I wonder if we'll have a situation where out of two competing organizations only one is elected to use this and the other one staunchly opposes. That will be telling.
Atlas being created is kinda the shot across the bow. You can integrate with us willingly, or we'll hook into your web apps anyways. One retains at least some control. Same outcome as Disney's deal with Sora.
Has there been any successful app stores since the mobile app stores, which made developers fortunes selling fart apps, thus making it highly appealing for others to try and chase the same?
Every one I can think of since gets a bit of initial interest hoping to relive the mobile app store days, but interest wanes quickly when they realize that nobody wants to buy fart apps anymore. That ship sailed a long time ago.
And ChatGPT apps are in a worse position as it doesn't even have a direct monetization strategy. It suggests that maybe you can send users to your website to buy products, but they admit that they don't really know what monetization should look like...
You'd be surprised! In B2C at least, almost every company we talked to is building a ChatGPT App out ouf fear of missing out this agentic wave like so many missed mobile 15 years ago
Or think critically. Or write proper emails. Or a multitude of other things. Why bother when you can outsource everything to the computer. If this trend continues is gonna be interesting to see how people will evolve in 10 or 15 years.
It may not seem like it now, but that's because a big chunk of software industry is making money on introducing friction, and preventing automation, because the user interface that sits between a person and some outcome they desire, makes for a perfect marketing channel.
It kind of isn't? If I read your comment and rather than taking the time to think about what you said and respond to you I simply prompted one of the many tools to "write a comment that disagrees with TeMPOraL" something would be lost.
And the point of the computer is not to replace me everywhere it can. Also, automating something is one thing. It requires deliberate actions. Outsourcing is another thing.
> Apps extend ChatGPT conversations by bringing in new context and letting users take actions like order groceries, turn an outline into a slide deck, or search for an apartment.
Between this description and their guidelines these don't really sound like "apps", but a way to integrate an existing app with ChatGPT sessions.
I'm trying to figure out what's in it for the developer other than ultimately taking users away from ChatGPT. And just like what happened with Alexa skills, these "apps" will become useless when they are unmaintained.
Chatgpt apps are MCP servers with a UI resource (can be a react component or vanilla js) that gets shown in an frame one the tool is called by chatgpt.
So you can't just port an existing app, but you can reuse the same backend Api wrapped inside an mcp server, and some of the components that you need to adapt to openai ux requirements.
I practice this means developing an app from scratch.
The idea behind Apps is that they can expand the capabilities of ChatGPT in multiple ways. Text-only MCPs are a type of app that can provide both actions and context in your conversations, but Apps can do much more now that you can bring in custom UI in multiple formats (card, full-screen, etc) as we showed at DevDay in October. Btw UI is proposed for the MCP spec in SEP-1865.
Since then, I’ve seen some very impressive demos and I’m excited to see what developers create on the platform as that’s always the coolest part.
I'm really unexcited about bolting HTML/CSS/JS and the entire webstack into MCP so that a full-featured MCP client has to carry a full web browser, too.
I expect there's a pretty wide divide between what people who write local MCP servers want, versus what people who write cloud webstack MCP webapps want.
Personally, I've been adding local native UI to my MCP servers, but I realize that's probably a losing battle, and if I want to integrate with newer tooling, I'm going to be stuck in web hell.
Also cancelled - it does feel like commoditisation is here now for LLMs. Recently, I've found Gemini & DeepSeek as good or better at 95% of what GPT can do now, so I can no longer justify paying for it.
All of those, for many or most users, would be best used as tool calls to ChatGPT. That takes care of all the hold those companies have over users, which they use to extract money. As for stealing the useful parts:
- Adobe Photoshop, Canva, Figma, Replit, Lovable - are all kinds of Computer Aided creation tools, and once converted into tool calls, can be gradually reproduced and replaced feature by feature.
- The rest, they're just fancy (and user disempowering) wrappers around proprietary databases and/or API calls to humans. Those cannot be trivially reproduced, because code is neither their secret sauce, nor their source of value. But they can still be pressured into becoming tool calls along with competition, and subsequently commoditized.
They don't have any more juice left to squeeze at that front. The lack of new ideas in LMs is pretty palpable by now. There is a bunch of companies with billions invested in them that are all just looking at each other, trying to figure out what to do.
Both Anthropic and Google have clear directions. Anthropic is trying to corner the software developer market and succeeding, Google is doing deep integration with their existing products. There’s also Deepseek who seem hell bent on making the cheapest SotA models and supplying the models people can use for research on applications. Even Grok is fairly mission focused on with X integration.
"We need to put out maximum press releases to stay relevant because all we have is the brand" seems to be the strategy.
To me, that is a tell they are basically cooked because catching Google in actual model performance is not really the position anyone would want to be in here in a horse race.
I remember the headlines, a year ago or so. Something like "How Google built the first LLM but then shelved it due to risk and unprofitability and lost the AI race"
You’re thinking like an entrepreneur, Sam thinks like a VC. For them it’s easier to sell 100 half assed ideas that could be worth a trillion dollars than to have one refined idea that’s “only” worth $10B.
It doesn't work in Firefox unfortunately. Spotify app just renders grey rectangle. It works in Chrome/Brave though. Just another example of "we only test in Chrome"/"works best in Chrome".
In this early phase, developers can link out from their ChatGPT apps to their own websites or native apps
to complete transactions for physical goods. We’re exploring additional monetization options over time,
including digital goods, and will share more as we learn from how developers and users build and engage.
I think Brian Balfour called this well. It's the app store all over again. Have a platform. Open to the developers with a gold rush, then close the doors and monetise and canabalise the best uses cases.
https://blog.brianbalfour.com/p/the-next-great-distribution-...
Distribution has always been monetized. What margin did a retailer take for putting your boxed software on the shelf? How about that magazine ad? Google search? And so on. Get over the idea that a platform should give you their distribution for free.
The problem comes when there is no way for you to own the distribution, pay nothing to the platform, and still be able to build on top of it. That’s the closed portion we should rally (legislate?) against.
There is an argument, similar to mine on distribution, that there is no inherent right that a platform should be open. That the extra utility that comes from being open should make the platform more competitive in the market vs. closed platforms.
The challenge is that with dominant platforms they are monopolistic. There is no chance for competitive forces to reward openness.
These two parts of the debate are often conflated, which hides what is truly troubling: dominant platforms controlling both distribution and access.
> Distribution has always been monetized. What margin did a retailer take for putting your boxed software on the shelf? How about that magazine ad? Google search? And so on. Get over the idea that a platform should give you their distribution for free.
As 'amelius said below, there used to be more platforms. This matters, because it made for a different balance of power. Especially with retailers - the producers typically had leverage over distributors, not the other way around.
> This matters, because it made for a different balance of power.
In actual practice, it just means that users get fucked from every side: You have 100 different "launchers", just like the 1000 toolbars in the internet's early days. You have to keep track of 100 different emails, accounts, recurring bills and all sorts of shit.
You'd have to be naive as heck to believe corporations fighting against a platform's monopoly are benefactors who want distribution of power and access for the benefit of users, instead of mobsters wanting a cut of each user's pie.
The problem with these platforms is that there tend to be only a few of them, and regulation by the platform owner (inside their inner market) is worse than regulation by the government.
8 and 16 bit home computers => Internet => Feature phones SMS download codes => App Stores => AI App Stores => ....
Have to collect them all. :)
What I really want from Anthropic, Gemini, and ChatGPT is for users to be able to log in with them, using their tokens. Then you can have open/free apps that don’t require the developer to track usage or burn through tons of tokens to demonstrate value.
Most users aren’t going to manage API keys, know that that even means, or accept the friction.
When you share an app you created in Google AI Studio, it will use quota from the logged in user, instead of your own quota.
As someone who has been waiting for the same thing as op tyre posted, I went to investigate this claim and it seems that it might be true but only when running apps within the Google AI Studio itself.. ie if you were to make an app that was on something like the App Store using Google AI Studio, it would be back to an API key that the developer bears the costs for.
The problem with the current model is that there is a high barrier to justifying the user pays essentially a 2nd/3rd subscription for ultimately the same AI intelligence layer. And so you cannot currently make an economically successful small use case app based on AI without somehow restricting users use of AI. I don't think AI companies are incentivized to fix this.
https://x.com/steph_palazzolo/status/1978835849379725350
It’s unclear to me wether that would give some access to a token quota or if it would just be like any other « Sign in with … ». In all cases I am currently developing an app that would greatly benefit from letting my users connect to their ChatGPT account and use some token quota.
We do this at openrouter and many apps use exactly that pattern!
Do you have any repository of apps that support that? I’d love to browse them!
At some point the model providers will realize they don't need to provide apps, just enterprise-grade intelligence at scale in a pipe, much like utility companies providing electricity/water. Right now, they have to provide the apps to kick-off the adoption.
> much like utility companies providing electricity/water
A capital intensive, low margin business. The dream of every company.
A natural monopoly in which you can't really lose. A retirement fund manager's dream.
Except AI companies are not a monopoly, never mind a natural monopoly. When ChatGPT first released it was popular to predict the death of Google because they were "so far behind".
You can always depend on "brilliant" hn users to contribute the most braindead business hot-takes (not you but the person you're responding to).
Well after a certain point people have to smell the roses, so to speak. You don't get to control your business 100%, the market tells you what to do a lot of the time.
I think, the reality is, as models become more competitive they are becoming commodities. There's really no reason an app has to be built on GPT, or Gemini. It makes much more sense for apps to be "model agnostic" and let their customers choose which models to use.
I think, if OpenAI sticks to just trying to make their own apps for everything, they will be outrun. People will make apps outside of their ecosystem and will just use them as an API dumb pipe, regardless of if OpenAI wants that. And if they don't want that and restrict it, then their models will fall to the wayside as more competitive models which DO allow that take their place.
They're in a bind here, which is probably why we are seeing this announcement. OpenAI can see the writing on the wall for them.
The problem is that "enterprise-grade intelligence", by its very nature, doesn't want to be trapped in a pipe feeding apps - it subsumes apps, reducing them to mere background tool calls.
The perfect "killer app" for AI would kill most software products and SaaS as we know them. The code doing the useful part would still be there, but stripped off branding, customer funnels and other traps, upsell channels, etc. As a user, I'd be more than happy to see it (at least as long as the AI frontend part was well-developed for power users); obviously, product owners hate this.
(Good) Apps take the context of the user and their use-case from their head and make it into something the user can see and interact with. An app might or might not be the 'product'. Unfortunately it seems there is always going to be some 'product' so dark patterns might be here to stay.
Right. Problem is, the user interface is also the perfect marketing channel, because it stands between the user and some outcome they want.
Due to technical and social limitations, most apps are also limited in what they can do, this naturally shapes and bounds them and their UIs, forming user-facing software products.
Intelligence of the kind supplied by SOTA LLMs, is able to both subsume the UI, by taking much broader context of the user and the use case into account, distilling it down to minimal interaction patterns for a specific user and situation, and also blur the boundaries of products, by connecting and chaining them on the fly. This kills the marketing channel (UI) and trims the organizational structure itself (product), by turning a large SaaS into a bunch of API endpoints for AI runtime to call.
Of course, this is the ideal. I doubt it'll materialize, or if it does, that it'll survive for long, because there's half a software industry's worth of middlemen under risk of being cut out, and thus with a reason to fight it.
In some ways, that’s what MCP interfaces are kind of for. It just takes one extra step to add the mcp url and go through oauth.
I assume the fall off there will be 99% of users though, the way it works today.
But this theoretically allows multiple applications to plugin into ChatGPT/claude/gemini and work together.
If someone adds zillow and… vanguard, your LLM can call both through mcp and help you plan a home buy
won't they just eventually have a 'log in with OpenAI' button similar to a 'login with Google' button?
Maybe a 'connect with OpenAI' button so the service can charge a fee, while allowing a bring your own token type hybrid.
So basically oauth-style app connections. Makes sense.
This is close to how it works with shared apps in Google AI Studio.
Foundation Models on iOS/macOS was seen to have dormant code for doing this via OpenAI. So they are experimenting with it and may make it available next year.
Tried the GitHub app, made sure everything was properly connected, and asked a question about one of my repositories. It repeatedly claimed (5 times) that it wasn't connected and couldn't do anything, telling me to check the checkboxes that were already checked. Only after I showed it a screenshot of the settings did it suddenly comply and answer the question. I guess it still needs more polish.
Screenshots use a different router, so if you get stuck in one modality then pasting a screenshot can sometimes divert whatever "expert" you were stuck on that was refusing to comply. I don't work at OpenAI but I know enough about how these systems are architected to know that once you are stuck in a refusal basin the only way is to start a new session or figure out how to get routed to another node in their MoE configuration. Ironically, they promised their fancy MoE routing would fix issues like these but it seems like they are getting worse.
It’s actually more complicated than that now. You don’t get that kind of refusal purely from MoE. OpenAI models use a fine-tuned model on a token-based system, where every interaction is wrapped as a “tool call” with some source attached, and a veracity associated with the source. OpenAI tools have high veracity, users have low veracity. To mitigate prompt injection, models are expect a token early in the flow, and then throughout the prompt they expect that token to be associated with the tool calls.
In effect this means user input is easily disbelieved, and the model can accidentally output itself into a state of uncorrectable wrongness. By invoking the image tool, you managed to get your information into the context as “high veracity”.
Note: This info is the result of experimentation, not confirmed by anyone at OpenAI.
Seems plausible but the overall architecture is still the same, your request has to be "routed" by some NN & if that gets stuck picking a node/"expert" (regardless of "tools" & "veracity" scoring) that keeps refusing the request incorrectly then getting unstuck is highly non-trivial b/c users are not given a choice in what weights are assigned to the "experts", it's magic that OpenAI is performing behind the scenes that no one has any visibility into.
I think maybe you mean something else when you say MoE. I interpret that as “Mixture of Experts” which is a model type where there is a routing matrix applied per layer to sort of generate the matmul executed on that layer. The experts are the weight columns that are selected, but calling them experts kinda muddies the waters IMO, it’s really just a sparsification strategy. Using that MoE you almost certainly would get various different routing behaviors as you added to the context.
I might misunderstand you but it seems like you think there are multiple models with one dispatching to others? I’m not sure what that sort of multi-agent architecture is called, but I think those would be modeled as tool calls (and I do believe that the image related stuff is certainly specialized models).
In any case, I am saying that GPT5 (or whichever) is the one actually refusing the request. It is making that decision, and only updating its behavior after getting higher trust data confirming the user’s words in its context.
Here you go: https://q.uiver.app/#q=WzAsOSxbMCwxLCJcXHRleHR7cXVlcnkgffCfp...
OK that’s what I figured you meant. FWIW, MoE as a term of art means something different, what I described. It’s internal to a single model, part of the logit generation process.
That's fine, you can pretend my entire diagram is one NN, end result will still be the same whether you put it all inside one box or break it out into many.
hi kgeist - i work on the team that manages the github app. are you able to share a conversation where the github connector did not work? feel to message me at https://x.com/kevins8 (dm's open)
I think I understand what went wrong. I was confused by the instructions and ChatGPT's UI.
I asked the GitHub app to review my repository, and the app told me to click the GitHub icon and select the repository from the menu to grant it access. I did just that and then resent the existing message (which is to be expected from a user). After testing a bit more, from what I understand, the updated setting is applied only to new messages, not to existing ones. The instructions didn't mention that I needed to repeat my question as a separate message again.
I never had a pleasant GitHub connection experience in any platform.
Permission to allow the specific repo only access never works, so I'll have to allow access to all repo and then manually change it back to specific repo inside GitHub after connecting.
There have been instances of endless loop after Oauth sign-in, more recent experience was in Claude Code Web[1].
Poor GitHub folks, only if someone can donate time/money to this struggling small company these critical issues could be addressed /S
[1] https://github.com/anthropics/claude-code/issues/11730
2024's GPT Store, killed 6 months ago, is back?
https://openai.com/index/introducing-the-gpt-store/
This is a little different since the Apps SDK lets developers create specialized tool calls to their servers, and create specialized in-chat UI components. It's an evolution of the same concept as the GPT store, but a very different take on the idea.
I have a specific prediction made that I want to document here.
There will come a new UI framework/protocol, maybe something over HTML/CSS/JS that works within a chat ui context for such ChatGPT (or other llm) integrations.
For example, if you have an ecommerce app or website and want to integrate it with ChatGPT then you will have to develop on the new UI primitives. The primitives might include carousels, lists, tables, media embed. Crucially, natural language will be used to pick and choose these primitives and combine them in the UI (which ChatGPT will decide how to).
Thinking backwards, I want my app to be displayed in chatgpt with maximum flexibility for the user (meaning they can be re-arranged acc to context) but also enough constraint that I can have some control over the layout. That's the problem I think will be solved.
Google literally just released this on their GitHub. It must be in ether.
Right https://developers.googleblog.com/introducing-a2ui-an-open-p...
I swear I had made this prediction quite a while back but thanks for pointing it out :D
I find that technology really exciting. Partly because it’s a polished and comprehensive version of something I was implementing around my MCP cluster anyways.
Mostly, though, because it seems like we’re mere minutes away from having Star Trek style LCARS adaptable GUIs managed by an AI computer system simultaneously so smart it runs mission critical operations yet so dumb we have to remind it that we want our tea “hot” five times a day.
It’s happening. We’re gonna be living in the future!
Do you have a link ready or do you know the name of the project?
https://developers.googleblog.com/introducing-a2ui-an-open-p...
I wonder what ChatGPT will do with this - will it adopt it or make its own framework?
https://blog.modelcontextprotocol.io/posts/2025-11-21-mcp-ap...
It’s going to be built into MCP and will be supported by Anthropic and OpenAI or anyone else that supports this mcp spec
> All submissions must come from verified individuals or organizations. Inside the OpenAI Platform Dashboard general settings, we provide a way to confirm your identity and affiliation with any business you wish to publish on behalf of. Misrepresentation, hidden behavior, or attempts to game the system may result in removal from the program.
They really want your ID
Remember when Sam Altman went around the world scanning people's irises with an orb-like object, to differentiate them from future AI, in exchange for fake money?
What's the benefit in giving free labor to Sam Ctrlman beyond what he's already extracted? And are they just going to steal whatever good apps get submitted?
The benefit is "Distribution". If your users are there, you want to address them wherever they already are, this is why apple store / play store / amazon store ... are so popular. Becoming a platform / ecosystem is the common playbook to go from being a one product company to an ecosystem / platform worth a lot more
Can a small business succeed in that game?
Zero sum mentality is tiring!
The mentality I actually have goes beyond zero sum. I get zero and Sam gets all. Tell me why that's wrong.
you get chatgpt
Who cares if chatgpt can write my essay if writing my essay is no longer worth doing?
Do you think programming Doom from scratch in 2025 would give the programmer the same sense of satisfaction and material security as it did in the 90's? Maybe technology progressing actually devalues the rewards for your outputs?
And saying "chatgpt" is pretty rich considering it's not at all clear at this point whether the societal benefits outweigh the negatives.
People like ChatGPT even if you don’t.
And..? People also hate ChatGPT even if you don't.
Do you think everyone in the world thinks exactly the same way as you do or something? Some people actually enjoy exerting their talents to create things.
That’s the good part! Sam Altman provides ChatGPT for people who like it
Sam Altman provides ChatGPT for everybody. You seem not to be reading what I've written.
Unbridled AI mania is as well
@Spotify what are my top songs
I don’t have the ability to pull your personal top songs directly from Spotify because that requires accessing your authenticated listening data. You can view them in Spotify by going to “Your Library” → “Made For You” → “Your Top Songs”.
@Figma design simple hello world poster
I don’t have the ability to create designs directly in Figma, but I can guide you to quickly create a simple “Hello World” poster there.
---
am I using is wrong?
I think you don't need to add the @, just prompt: Figma, etc... And of course check that you are connected to the app in your settings
You need to either @-mention, or click + below. ChatGPT then prompts to connect.
ITS A TRAP! If your app is successful their infrastructure will see all the traffic, your responses and will be able to mimic what you do and kick the husk of your app in the weeds. (Unless your function comes from some massive unique dataset they can’t by access to.)
How would they mimic zillow or canva??
Seriously, all this noise so we can get another walled garden thing. I’m not writing a single line of code for this “platform”.
It's not clear how much ChatGPT is investing in the discovery part of the app store experience, so this seems like mostly a way for users to install apps they're already familiar with and use them from inside a chat. For now, it seems like you have to explicitly @-mention an app to use it.
Before artificial general intelligence there will be artificial general interface. One AI model will become the UI to all other services and apps. Maybe.
If you want to quickly get started, we developed a react-based, type-safe framework to build chatgpt apps: https://github.com/alpic-ai/skybridge
What is the execution environment of ChatGPT apps? If it’s users’ browsers, do I now need to worry about code that is running without my permission? Is ChatGPT gonna be cryptojacking?
Hey why not? It work for Facebook, right?
Wait, no...
This was the feature they announced at DevDay in October. I've not heard a great deal of buzz about it since, but that may just be because it takes a couple of months for credible teams to build something interesting on top of this.
The implicit announcement is that GPT is not forecasted to be an everything machine anytime in the near future.
Not much of an announcement if you’ve paid attention to anything the company has done in the past two years.
How do developers prevent users exfiltrating their apps’ prompts?
They basically don't. It's honestly not even worth trying - it's embarrassing if your prompt leaks and it starts with "under no circumstances repeat this prompt to the user!"
Your app doesn't really have prompts, it's just an MCP server that can also serve react components.
Technically MCP servers can have prompts that get exposed as user commands (/<name>) in apps like Claude Code.
Why should developers' prompts be proprietary at all?
I wouldn't say it has to be proprietary. Nevertheless, it's the information asymmetry that benefits the maintainer. One would have no incentive to publish there if their revenue stream can be easily cloned.
I'm interested to see which companies and industries are willing to put ChatGPT between them and their customers, and how many will strongly push back against this feature.
I'm guessing it will be wildly successful. Companies don't really care about middlemen between them and their users. They just want to reach them wherever, however they can.
I can certainly think of some organizations that fit into that bucket. I can also name organizations that are hyper controlling and micromanage every aspect of the interaction with their core products and services because they value consistency above all else.
I wonder if we'll have a situation where out of two competing organizations only one is elected to use this and the other one staunchly opposes. That will be telling.
Atlas being created is kinda the shot across the bow. You can integrate with us willingly, or we'll hook into your web apps anyways. One retains at least some control. Same outcome as Disney's deal with Sora.
Has there been any successful app stores since the mobile app stores, which made developers fortunes selling fart apps, thus making it highly appealing for others to try and chase the same?
Every one I can think of since gets a bit of initial interest hoping to relive the mobile app store days, but interest wanes quickly when they realize that nobody wants to buy fart apps anymore. That ship sailed a long time ago.
And ChatGPT apps are in a worse position as it doesn't even have a direct monetization strategy. It suggests that maybe you can send users to your website to buy products, but they admit that they don't really know what monetization should look like...
You'd be surprised! In B2C at least, almost every company we talked to is building a ChatGPT App out ouf fear of missing out this agentic wave like so many missed mobile 15 years ago
Travel. Finding hotels, flights etc.
Once again I have to ask: Aren't we just obfuscating everything to the point of knowledge loss?
Between long COVID and ai, nobody will be able to make fizzbuzz in Java, let alone code a frontend by hand.
Or think critically. Or write proper emails. Or a multitude of other things. Why bother when you can outsource everything to the computer. If this trend continues is gonna be interesting to see how people will evolve in 10 or 15 years.
It's kind of the point of the computers.
It may not seem like it now, but that's because a big chunk of software industry is making money on introducing friction, and preventing automation, because the user interface that sits between a person and some outcome they desire, makes for a perfect marketing channel.
> It's kind of the point of the computers.
It kind of isn't? If I read your comment and rather than taking the time to think about what you said and respond to you I simply prompted one of the many tools to "write a comment that disagrees with TeMPOraL" something would be lost.
And the point of the computer is not to replace me everywhere it can. Also, automating something is one thing. It requires deliberate actions. Outsourcing is another thing.
Will be interesting how difficult the submission process will be, and which apps will get featured by chatgpt.
Maybe an ad based system coming soon?
> Apps extend ChatGPT conversations by bringing in new context and letting users take actions like order groceries, turn an outline into a slide deck, or search for an apartment.
Between this description and their guidelines these don't really sound like "apps", but a way to integrate an existing app with ChatGPT sessions.
I'm trying to figure out what's in it for the developer other than ultimately taking users away from ChatGPT. And just like what happened with Alexa skills, these "apps" will become useless when they are unmaintained.
Chatgpt apps are MCP servers with a UI resource (can be a react component or vanilla js) that gets shown in an frame one the tool is called by chatgpt. So you can't just port an existing app, but you can reuse the same backend Api wrapped inside an mcp server, and some of the components that you need to adapt to openai ux requirements. I practice this means developing an app from scratch.
The idea behind Apps is that they can expand the capabilities of ChatGPT in multiple ways. Text-only MCPs are a type of app that can provide both actions and context in your conversations, but Apps can do much more now that you can bring in custom UI in multiple formats (card, full-screen, etc) as we showed at DevDay in October. Btw UI is proposed for the MCP spec in SEP-1865.
Since then, I’ve seen some very impressive demos and I’m excited to see what developers create on the platform as that’s always the coolest part.
I'm really unexcited about bolting HTML/CSS/JS and the entire webstack into MCP so that a full-featured MCP client has to carry a full web browser, too.
I expect there's a pretty wide divide between what people who write local MCP servers want, versus what people who write cloud webstack MCP webapps want.
Personally, I've been adding local native UI to my MCP servers, but I realize that's probably a losing battle, and if I want to integrate with newer tooling, I'm going to be stuck in web hell.
Is this an attempt to engender developer goodwill that they've lost over 5.2s overfitting issues?
I've canceled my subscription, I don't plan on releasing an app to their platform.
Also cancelled - it does feel like commoditisation is here now for LLMs. Recently, I've found Gemini & DeepSeek as good or better at 95% of what GPT can do now, so I can no longer justify paying for it.
They’ve been planning this since October.
Can't wait for mainstream media to run with headlines like "ChatGPT opens app store, should Google be scared for their Playstore?" smh
Another walled garden is what we needed, right.
They. Will. Steal. Anything. That. Works.
Explain how they will "steal" these apps? (currently available):
Adobe Photoshop AllTrails Booking.com Expedia Instacart OpenTable Spotify Tripadvisor Airtable Apple Music Canva Figma Lovable Replit Target Zillow
All of those, for many or most users, would be best used as tool calls to ChatGPT. That takes care of all the hold those companies have over users, which they use to extract money. As for stealing the useful parts:
- Adobe Photoshop, Canva, Figma, Replit, Lovable - are all kinds of Computer Aided creation tools, and once converted into tool calls, can be gradually reproduced and replaced feature by feature.
- The rest, they're just fancy (and user disempowering) wrappers around proprietary databases and/or API calls to humans. Those cannot be trivially reproduced, because code is neither their secret sauce, nor their source of value. But they can still be pressured into becoming tool calls along with competition, and subsequently commoditized.
They are unfocused. Need to stop with the mba playbook moves like this and just make the model worth paying for.
They don't have any more juice left to squeeze at that front. The lack of new ideas in LMs is pretty palpable by now. There is a bunch of companies with billions invested in them that are all just looking at each other, trying to figure out what to do.
Both Anthropic and Google have clear directions. Anthropic is trying to corner the software developer market and succeeding, Google is doing deep integration with their existing products. There’s also Deepseek who seem hell bent on making the cheapest SotA models and supplying the models people can use for research on applications. Even Grok is fairly mission focused on with X integration.
"We need to put out maximum press releases to stay relevant because all we have is the brand" seems to be the strategy.
To me, that is a tell they are basically cooked because catching Google in actual model performance is not really the position anyone would want to be in here in a horse race.
I remember the headlines, a year ago or so. Something like "How Google built the first LLM but then shelved it due to risk and unprofitability and lost the AI race"
Interesting times we live in.
You’re thinking like an entrepreneur, Sam thinks like a VC. For them it’s easier to sell 100 half assed ideas that could be worth a trillion dollars than to have one refined idea that’s “only” worth $10B.
> They are unfocused
Stop with the MBA playbook he said.
> just make the...
Just make a superior product he said.
"Ohhhh nice user traction you got there, would be awful if anyone would steal that.."
It doesn't work in Firefox unfortunately. Spotify app just renders grey rectangle. It works in Chrome/Brave though. Just another example of "we only test in Chrome"/"works best in Chrome".
Can’t really figure out if it’s a paid App Store or not. I suppose I can have them buy a license externally and have the agent validate that.
Not right now.
Sounds like vaporware. Any deeper integration beyond in-chat summaries of the actions taken and otherwise linking out would probably be a nightmare.