Late model: OpenAI GPT Store may debut next week

OpenAI’s GPT Store – a one-stop shop for customized chatbot models – is expected to start business next week, after missing its planned debut last month amid boardroom turmoil at the startup.

OpenAI teased its store and detail-free revenue-sharing plan in early November during its first developer conference. The idea is that developers can create AI chatbots tuned for specific purposes, using a low-code process that’s more configuration than programming, and distribute these apps, called GPTs, via OpenAI’s souk to customers in the hope of getting paid.

The opening of the GPT Store was delayed, however, when later that month the super lab fired CEO Sam Altman, which alarmed major investor Microsoft, and ultimately led to Altman’s reinstatement. With the executive shuffle and takeback complete, OpenAI’s app store venture appears to be back on track.

A Hacker News account that purports to belong to the CTO and co-founder of Xip, Eddie Siegel, reports receiving an email from OpenAI about the debut of the GPT Store next week. Siegel could not immediately be reached to confirm this. But others posting to the forum claimed they too had received the store announcement email.

Artificial intelligence is a liability
MUST READ

Separately, The Information has published a report that states the same. OpenAI did not respond to a request for comment.

The initial description of the GPT Store indicated that verified builders would have the opportunity to create models focused on specific markets and OpenAI would provide discoverability and monetization infrastructure. It would also promote GPTs deemed to be “the most useful and delightful” in categories like productivity, education, and “just for fun.”

To sweeten the pot, OpenAI announced that it will indemnify enterprise and developer customers from potential copyright claims – a shadow that now hangs over the entire AI industry on account of its penchant for training models on audio, video, text, and images without permission from authors.

Indemnification will become more meaningful for developers if and when they start earning money from GPT models. Revenue tends to motivate legal scrutiny, after all.

Two examples of GPTs have already been revealed. One links to graphics tool Canva, allowing chat-based commands to drive graphics creation. Another one links to automation service Zapier to enable app automation. This might involve, for example, creating a chat-based assistant to help with schedule management in Google Calendar.

What the AI copyright fights are truly about: Human labor versus endless machines
PODCAST

Developers have been working on complementary services. Gapier, for example, is trying pre-configured ‘actions’ that can be added to GPTs – much as an NPM package might be added to a web app – to perform specific functions, like conducting a Google Search or generating an image from text.

OpenAI also expects some developers will look beyond the low-code approach by defining custom actions that make a broader set of APIs available to GPTs.

“Connect GPTs to databases, plug them into emails, or make them your shopping assistant,” OpenAI suggests. “For example, you could integrate a travel listings database, connect a user’s email inbox, or facilitate e-commerce orders.”

And this should pose no risk at all, given a chatbot that always gets things right and can’t be manipulated through adversarial prompting. Sarcasm filters fully on for that last statement.

Essentially, the GPT Store is an effort to get developers to integrate OpenAI’s paid APIs into apps and to pass that cost on to customers. Developers considering the possible advantages and disadvantages of betting on OpenAI’s platform – sure to face competition from the lively open source LLM community – are eager to see the financial details before they render a verdict. ®