OpenAI's Dev Day Announcements Target the Traditional App Store Model
At its Dev Day event on Tuesday, OpenAI's headline items were its agentic assistants called Dots and fresh AI models. Viewed as a whole, however, the announcements add up to a larger strategy: an attempt to unsettle the established app store model. Together, the releases make ChatGPT the venue where software is found, started and used by humans and agents alike.
OpenAI also unveiled a mechanism that lets individuals carry their ChatGPT identity across services, and apply the AI allowance they already have inside third-party applications.
Experimenting with how apps might live inside its well-known chatbot interface is not new for OpenAI, but the company's present approach looks considerably more developed than earlier attempts.
To begin with, OpenAI is making ChatGPT itself a surface from which apps are launched. The chatbot, which the company says reaches 1.2 billion weekly users, has not yet made the most of its capacity to serve as a discovery route for locating and using apps that work with AI.
To address that, ChatGPT will start proposing apps inside the flow of a conversation when it detects that a given app might help the user finish a task. The user can then link the app and start using it straight from ChatGPT.
An expanded plugin architecture, which now accommodates extensions, also supports this. Developers are able to create interactive panels where users can operate their tools while chatting with ChatGPT. In practice, this converts apps and services that users previously reached through the web or a native desktop or mobile app into something run from inside ChatGPT.
Developers who join the system can produce AI-native versions of their apps through ChatGPT, in the same manner they would for the open web or a mobile app store. With a growing share of discovery happening via AI chat, it becomes a distribution channel that is difficult to pass up.
Users are given a reason to use that channel as well, since "Sign in with ChatGPT" will allow them to bring their AI allowance along. OpenAI has 16 launch partners on this effort, among them Cognition's Devin, Notion, Vercel, T3, OpenClaw, and Dactyl, and says more will follow soon.
During a demonstration at OpenAI's Dev Day event, the company showed how its own new meeting app could operate inside ChatGPT, displaying upcoming meetings from the user's calendar. From there, the user could readily opt to have AI take meeting notes, and then get a summary of follow-up items once the meeting ended.
In a further example, users would work with design-focused apps such as those from Figma and Adobe to revise their current project or use a particular feature that would otherwise have demanded a standalone app.
The apps can be shared with other people, such as work colleagues, through the lightweight websites ChatGPT now generates. From these ChatGPT sites, a user's coworkers could log in to the app using their own credentials and permissions, which makes the software experience personalised to each of them.
OpenAI also discussed enhancements to the way developers submit plugins for review — its counterpart to Apple's App Review process. Developers will now be able to follow their review, see what needs correcting, ask for a human review, and update the tools in their plugin without restarting the entire submission.
The company additionally announced a new enterprise app marketplace where partners can promote their services. At launch, more than 30 partners are offering their app through the OpenAI Marketplace, among them Adobe, Figma, Sierra, Decagon, HubSpot, Salesforce, ServiceNow, Harvey, Legora, Palo Alto Networks, CrowdStrike, Baseten, and others. Eligible customers can put part of their OpenAI commitment toward approved partner software, OpenAI said.
Beyond making ChatGPT an app discovery tool where users connect with apps directly, OpenAI's embrace of agents is gently steering users into a new period in which they need not think as much about which app to use or why, instead depending on suggestions from their AI agents.
With the launch of its autonomous AI agents known as Dots, agents will browse the web on a user's behalf, carrying out necessary tasks within their own cloud infrastructure. Every Dot has its own cloud computer and browser, and is also able to use connected apps. Under the old model, a user who wanted to manage some kind of project — building a website or coordinating a meeting calendar, for instance — had to download various apps onto their computer or phone to make it happen.
Now, instead, users will tell an AI agent what they want done, and it will simply carry it out. In certain cases, the agent's work may call for access to another AI model or a specific application. But that is something the agent may bring to the user, rather than the reverse — where the user is the one who goes out to find the particular tool for the job.
As OpenAI explains, the Dots will be able to connect to its ecosystem of now over 4,000 apps.
What is more, the Dots can be proactive, perhaps stepping in to offer help with a project or bug, or handling a task the user forgot — though all of this is done with the user's approval. The Dots will perform what OpenAI calls "proactive research" in the background using connected apps in a read-only mode, but users would still retain control over the actions taken.
This system could give app makers an additional way to reach customers outside Apple and Google's app stores. For instance, users could subscribe to a service through the app maker's website, then link that account to ChatGPT or Codex for their agent to use. OpenAI has not said, however, that these features would remove the need for separate third-party subscriptions or other charges.
Notably, OpenAI did not tease any plans on this front on Tuesday, concentrating instead on how it is building the discovery, distribution, and interface pieces to counter the app store model, along with an identity layer and a mechanism for sharing a user's existing ChatGPT AI allowance. It announced no plans on Tuesday for a billing or revenue-sharing system comparable to the economic layer of traditional app stores.
to like, bookmark, and comment.
Comments
Join the conversation:
No comments yet. Be the first to share your thoughts.