- OpenAI's DevDay announcements, as TechCrunch reports them, make ChatGPT a place to discover, launch and use apps, with sign-in and agents attached.
- The pieces announced cover discovery, identity and AI allowance. TechCrunch says OpenAI announced no billing or revenue-sharing system comparable to an app store's.
- Leaders should ask who controls sign-in, permissions and plug-in updates for any ChatGPT app their staff use.
A store needs three things
An app store does three jobs. It helps people find software. It proves who they are. It takes their money. At DevDay, OpenAI showed pieces of the first two jobs. TechCrunch reports that the company did not announce the third.
That is the useful way to read the news. The question for a business is not whether ChatGPT replaces Apple's or Google's store. The question is who will hold each of the three jobs, and what that means for the software your staff already use.
What OpenAI announced
TechCrunch reports that ChatGPT will suggest apps inside a conversation when it judges one could help with the task. The user can connect the app and use it without leaving the chat.
The mechanism is an expanded plug-in system that now supports extensions. Developers can build interactive panels that sit beside the conversation. Work that once happened on a website or in a desktop or mobile app can now happen in that panel.
Sign-in is the second piece. "Sign in with ChatGPT" lets people bring their ChatGPT identity and their existing AI allowance into third-party apps. OpenAI named 16 launch partners, including Notion and Vercel.
Sharing is the third. Apps can be passed to colleagues through lightweight websites that ChatGPT now produces. Co-workers sign in with their own credentials and permissions, so each person sees a version set up for them.
Software that comes to the user
The agents may matter more than the panels. OpenAI's new agents, called Dots, do their work on OpenAI's own cloud infrastructure. According to TechCrunch, each Dot has its own cloud-hosted computer and browser, and it can also draw on apps the user has connected.
This changes who picks the tool. Today a person decides which app to open. With an agent, the agent may bring the tool to the person. The buyer of software then has a new audience to win: the agent's suggestions.
OpenAI says Dots can be proactive, but only with the user's approval. Their background research runs on connected apps in read-only mode. Those limits are OpenAI's own description. TechCrunch did not report any independent test of them.
The cash register is missing
TechCrunch is explicit on billing. OpenAI did not announce plans on Tuesday for a system to bill customers or share revenue the way established app stores do. It also has not said these features remove the need for separate subscriptions or other charges.
TechCrunch offers one example of how money could still flow. A customer could buy a subscription directly from an app maker's site, then link that account to ChatGPT or Codex for the agent to use. For app makers, that is one more route to customers outside Apple's and Google's stores.
For enterprises, OpenAI added a marketplace. TechCrunch lists some 30-plus launch partners, including Salesforce, ServiceNow, HubSpot, Palo Alto Networks and CrowdStrike. OpenAI said that qualifying customers may put some of their spending commitment with OpenAI toward approved partner products.
What this means for risk
This is one announcement, not a measured shift. TechCrunch's reading is that it points toward disruption of the app store model. Adoption data was not part of the report.
Still, the design has a consequence for security and procurement. When identity, permissions and the agent's browser all sit with one vendor, that vendor becomes part of every connected app's trust chain. This is the publication's analysis, not a finding in the report.
The review process shows the same dependence. OpenAI says developers can now update a plug-in's tools without restarting their whole submission. That helps developers. Buyers should ask what review the updated tools receive.
Questions to put to your teams
First, which ChatGPT apps are staff already using, and who approved them? Second, when a colleague opens a shared ChatGPT site, whose permissions apply, and who can see them? Third, what do the marketplace terms say about the commitment you may apply to partner software?
Fourth, if an agent suggests a tool, who owns that purchase decision? Fifth, where do you pay: through OpenAI, or through the vendor's own site? The report says OpenAI has not announced how that will work.
A store without a cash register is still a front door. In the publication's view, the company that controls the door and the ID desk could shape how software is chosen, whoever ends up taking the payment.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: TechCrunch.
