Maybe Not Everything Needs an App Store
Stephanie Palazzolo
You get why everyone wants to build an "app store" – everyone wants to be a platform. And the way that Microsoft, with Windows, Apple, with iOS, and Google, with Android, have been able to do this is via applications. And in the era of mobile, that means app stores. But what if, just bear with me, everything isn't meant to have an app store? What if new technology requires new ideas?1
Add to this the fact that this particular technology is evolving so quickly that it feels like trying to have an underlying platform, in the traditional sense, is going to be a challenge for a while. The iPhone worked, in part, because it was technology people intuitively understood how to use. You ran applications just like you do on a regular computer, just on your phone. The main difference (beyond screen size) was the touch interface. But even that was akin to a mouse, just more natural (while being less precise).
ChatGPT is something different. Sure, chatbots have existed for years, but that chat element of the service is arguably the least interesting aspect of it. Again, it's about the technology underneath. If anything, the chat element can sometimes get in the way and you're going to see it become just one way you can interact with the underlying AI. So asking developers to build chatbots just seems like a temporary opportunity at best. And apparently it's not even really working from the get-go.
However, Kirilenko and four other GPT Store developers said fewer customers have used their chatbots than they expected. Demochkin said around 5% of the more than 36,000 custom chatbots he analyzed received 150 to 500 active users per day, but the vast majority only landed one or two users per day, if that.
Part of this is undoubtedly because the capability is behind the $20/month premium paywall. And I suspect OpenAI will have to lift that in an effort to spur usage. But I also suspect even that won't work beyond some initial novelty.
Another part is just that OpenAI is doing so much at the moment. There's not a singular focus on making this paradigm work – including apparently, combating spam. Again, it's just too early, with everything moving too fast, and not enough of an establish framework for how this should all work. Everyone is figuring it out from OpenAI on down.
The struggles of the GPT Store and the lack of attention its leaders have given to it may also be a consequence of the company’s dizzying array of projects and priorities. They include upcoming products such as a search engine, AI agents, and Sora AI–generated video, plus the development of its next core large language model and Altman’s ambitious AI chip venture to aid OpenAI.
I don't know the answer, but my sense is that they have to think about being a platform differently than what has worked for the last generations of computing.
1 This includes Apple's own newer products too. It's not like third-party apps on the Apple Watch have exploded in popularity. The jury is still out with Vision Pro, but I suspect it's mainly going to be an entertainment device for quite some time. And that's fine. Not everything needs to be a platform from day one!