OpenAI introduces the GPT Store

First announced back at their dev days, today OpenAI officially opened up their app store. I’m looking at how it’s structured, what applications are featured, and the missing payouts. I’m still waiting for access to the store to be rolled out to me, so starting with the glaring hole: there is still no transparency on the revenue sharing. OpenAI briefly noted:

In Q1 we will launch a GPT builder revenue program. As a first step, US builders will be paid based on user engagement with their GPTs. We’ll provide details on the criteria for payments as we get closer.

It doesn’t inspire confidence as a small builder. Larger players can cross-subsidize development of their GPTs to capture customers and keep up on the potential.

a few hours later Took a minute, but I’m in the store. First impression is the ChatGPT store is bog standard - a weekly featured section followed by trending applications across categories. As for what made the cut: the featured section has 4 apps, 3 of which are from large companies (AllTrails, Khan Academy, Consensus). There is one “indie” application, “Booksight”, which aims to help the user pick a book. It’s alright, but didn’t blow me away compared to plain ChatGPT.

The individual categories each show the six above-the-fold applications that’re currently trending. That’s about it for app-store structure, so time for hot-takes on what’s trending 🍿

I’m tickled that the top applications in the ‘Writing section are to humanize some text, e.g., cover up that some text was written by an LLM. Another curious application (veed) is to produce video from prompts. I like the idea but it used stock footage of a pregnant belly alongside voice-over to produce a surreal & baffling mash-up when I asked for a penguin surfing. I’m crying.

The programming section is dominated by tools to create websites - a particularly interesting tool reverse engineers a website from an image. There’s also an educational ‘AskToCode’ GPT which uses source code from GitHub to augment the prompt and answer questions. This app is also an example of partnering with against an identity provider, in this case GitHub.

Overall, some interesting applications, but nothing which truly blew me away relative to the underlying GPT-4 capabilities, e.g., Khan’s tutor provided an excellent description of eigenvectors, but I’d bet I could get 90% of the way there with a bit of prompting on top of GPT-4. This might mean that there’s still room to get a wedge in via a GPT that is unique and stands out.

And that brings me to a tangential idea that I can’t get out of my head: a sort of power OCR for handwritten text. I’ve brought up OCR for my handwriting before but The Power is handwriting plus edits and markup. For example, I haven’t had success with underlines, or strikethroughs, or italics. How about emojis 😂🐶🍔? A caret to insert ^ text maybe? (editor: it didn’t). This “Super OCR” might not have general appeal, but at least one person would be impressed (that’s me!).