The focus of OpenAI’s Dev Day on Tuesday may have been on agent assistants known as Dots and new AI models, but taken together, the AI company’s announcements hint at a bigger plan: disrupting the traditional app store model. In summary, today’s announcement transforms ChatGPT itself into a place where people and agents alike can discover, launch, and use software.
Additionally, OpenAI introduced a way to carry your ChatGPT ID with you while allowing you to use your existing AI allowances with third-party apps.
This isn’t the first time OpenAI has experimented with how apps can interact within a familiar chatbot interface, but the current vision feels more fleshed out than before.
First, the company is turning ChatGPT itself into a surface for launching apps. Chatbots currently have 1.2 billion weekly users, according to the company, but their potential as a discovery mechanism for finding and using apps that work with AI has yet to be tapped.
To change this, ChatGPT will start suggesting apps in the flow of the conversation if it recognizes that a particular app will help the user complete a task. From there, users can connect to the app and use it directly within ChatGPT.
This is also aided by enhancements to ChatGPT’s plugin architecture, which now supports extensions.
This allows app developers to build interactive panels that allow users to interact with tools while chatting with ChatGPT. This essentially changes the apps and services that users previously used via the web or through native desktop or mobile apps to those that are interacted with directly within ChatGPT.
Developers who sign on to the system can build AI-native versions of their apps through ChatGPT in the same way they use the open web or mobile app stores. With more and more being discovered through AI chat, this is a distribution channel that cannot be overlooked.
Users also get an incentive to use that channel by using Sign in with ChatGPT, which brings in AI quotas. (OpenAI has 16 launch partners participating in the effort, including Cognition’s Devin, Notion, Vercel, T3, OpenClaw, and Dactyl, and the company says it plans to add more soon.)
In a demo at OpenAI’s Dev Day event, the company showed off how its new conferencing app works within ChatGPT and can display upcoming meetings from a user’s calendar. Here, users can easily choose to use AI to take meeting notes and receive a summary of follow-up items when the meeting ends.
In another example, a user might use a design-focused app like Figma or Adobe to work on a revision of a current project or use a specific feature that would have required a standalone app.
ChatGPT creates a lightweight website that allows you to share your apps with others, including your work colleagues. From these ChatGPT sites, your colleagues can sign in to the app using their own credentials and permissions, allowing them to personalize their software experience.
OpenAI also talked about improvements to the way developers submit plugins for review. This is the OpenAI version of Apple’s app review process. Now, developers will be able to track reviews, see what needs fixing, request human reviews, and update the plugin’s tools without having to redo the entire post.
The company also announced a new enterprise app marketplace where partners can market their services. At launch, more than 30 partners are offering apps through the OpenAI Marketplace, including Adobe, Figma, Sierra, Decagon, HubSpot, Salesforce, ServiceNow, Harvey, Legora, Palo Alto Networks, CrowdStrike, and Baseten. OpenAI says eligible customers can apply some of the OpenAI commitments to approved partner software.
In addition to turning ChatGPT into an app discovery tool that connects users directly to apps, the adoption of OpenAI’s agents is gently nudging us into a new era where users no longer have to think so much about which apps they use and why, instead relying on AI agent suggestions.
By launching autonomous AI agents, known as Dots, the agents navigate the web on your behalf and perform tasks you need to perform on your own cloud infrastructure. Each Dot has its own cloud computer and browser, and can also use connected apps. Compare this to the older model. Users who wanted to manage some kind of project, such as building a website or coordinating a meeting calendar, had to download various apps to their computers or mobile phones to make it happen.
Now, users can simply tell an AI agent what they want the agent to do for them, and the agent will do it for them. In some cases, an agent’s work may require access to another AI model or a specific application. But this is what the agent brings to the user, rather than the other way around, which causes the user to go looking for the specific tools they need to do their job.
As OpenAI explains, Dots will now be able to connect to its ecosystem of over 4,000 apps.
Additionally, Dots can be proactive, offering help with projects, bugs, and even taking care of tasks you forgot, all with your permission. Dots uses connected apps in read-only mode to perform what OpenAI calls “proactive research” in the background, but users still have control over their actions.
The system could give app makers another way to reach customers outside of Apple and Google’s app stores. For example, users can register for the service through the app maker’s website and connect their account to ChatGPT or Codex for use by agents. However, OpenAI does not state that these features eliminate the need for separate third-party subscriptions or other fees.
Notably, OpenAI did not reveal any plans on this front today, instead focusing on how it is building out the discovery, distribution, and interface pieces to compete with the app store model, along with the identity layer and mechanism for sharing users’ existing ChatGPT AI quotas. No plans were announced Tuesday for a billing or revenue-sharing system comparable to traditional app store economic tiers.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
