When OpenAI’s Sam Altman took the stage at Dev Day 2025 in San Francisco, the demos looked, at first, like any other product update: new models, cheaper APIs, faster responses. But then something unusual happened: the announcement that OpenAI wants ChatGPT to be more than a chatbot. The company is positioning it as the next portal for the internet — a place where apps, agents and content live together in a single thread, blurring the boundaries between browser, operating system and assistant.
In a post on its website, OpenAI said users can now run, chat with and build apps directly in ChatGPT powered by a new Apps SDK that allows developers to embed interactive software into the chat window itself. At launch, it partnered with companies like Canva, Figma, Spotify, Coursera and Zillow, letting users design graphics, listen to music, take courses or browse property listings without ever leaving ChatGPT.
The move makes the AI chatbot a “container” of sorts for the wider web — where users can type or speak requests, and apps respond contextually inside the same conversation.
The Browser That Talks Back
This shift builds on a broader trend: The web’s dominant interface has long been visual, and users navigate by clicking through apps and tabs. OpenAI’s new framework may flip that model, making conversation itself the interface.
The new AgentKit is another fascinating piece. Announced alongside the app integrations, it gives developers a drag-and-drop workflow builder to create agents that can execute tasks, connect with APIs or integrate with enterprise systems. It’s essentially OpenAI’s answer to workflow tools like Zapier or Lindy AI but embedded natively inside ChatGPT.
MORE FOR YOU
In its announcement, the $500 billion startup said developers can create complex, goal-oriented agents using a mix of visual blocks and Python scripting.
“Until now, building agents meant juggling fragmented tools — complex orchestration with no versioning, custom connectors, manual eval pipelines, prompt tuning, and weeks of frontend work before launch,” the post notes. “With AgentKit, developers can now design workflows visually and embed agentic UIs faster using new building blocks.”
The tool includes embeddable chat components for websites and evaluation systems to test an agent’s accuracy.
These updates position OpenAI as a consumer platform as well as a developer ecosystem — two roles typically filled by Apple’s iOS and Google’s Android.
Living Inside ChatGPT
From a consumer standpoint, OpenAI’s direction hints at the consolidation of online behavior into a single conversational interface.
For years, users have often toggled between a fragmented universe of apps — Slack for work, Spotify for music, Chrome for browsing, Figma for design. ChatGPT’s new structure may complicate that complexity: In theory, a user could brainstorm a logo, design it in Canva, check licensing terms, and share it — all inside the same chat.
That’s an enticing vision for convenience, but it also raises several questions — for instance, about control. If ChatGPT becomes the environment where users “live,” OpenAI becomes the intermediary for everything they do online.
Furthermore, as Wired noted, this is not OpenAI’s first effort to introduce ChatGPT apps – the company announced a way to build custom widgets or GPTs at its developer conference two years ago, but it did not prove to be a big hit.
This also positions OpenAI directly against the tech giants that dominate digital access today. Apple’s iOS, Google’s Android, and even Microsoft’s Windows rely on graphical navigation and native apps. OpenAI’s conversational OS, by contrast, could be swept under or undercut that hierarchy.
“OpenAI is competing with Google, Anthropic, and Amazon in trying to lure developers to its services,” notes Wired. “It is also under pressure as rivals like Meta and DeepSeek launch capable open source models, which allow developers to fine-tune premier models for their needs without having to pay for access to OpenAI’s APIs.”
Risky Bets
Alongside Apps SDK and AgentKit, OpenAI opened up its most advanced models to developers, including GPT-5 Pro for complex reasoning and Sora 2, its video-generation model, through expanded API access. It also rolled out GPT-5 Codex, the latest iteration of its code-generation model, now generally available with new Slack integrations and administrative controls for enterprise deployment.
Are these an attempt to turn OpenAI’s technology stack into the infrastructure layer for both consumer and enterprise AI? The more developers build apps and agents inside ChatGPT, the more entrenched OpenAI becomes as the default environment for interaction — perhaps much like browsers became the default gateway for the web two decades ago.
However, the comparison cuts both ways: If ChatGPT really is becoming a new browser, it may inherit the same scrutiny browsers faced, and for good cause — around privacy, interoperability, and gatekeeping.
Also, embedding third-party apps and agents directly into ChatGPT could introduce entirely new vectors for privacy and data risk. Users will have to trust that embedded apps handle data securely and that OpenAI’s review system keeps malicious agents out.
There’s also the question of lock-in. Developers building for ChatGPT’s ecosystem may find themselves constrained by its policies or monetization rules — similar to how mobile developers rely on Apple’s App Store and Google Play.
Regulators may continue to be wary, too. A conversational platform that hosts apps, monetizes transactions, and intermediates user data could fall under new forms of antitrust and digital-market oversight. The European Union’s Digital Markets Act (DMA) has already signaled tighter scrutiny for “gatekeeper platforms.”
There’s also the crucial human choice factor: will users actually want to stay inside ChatGPT all day? The idea of a single conversational layer running your digital life sounds efficient — until we note how people use devices socially, visually and emotionally.
Nick Turley, OpenAI’s head of product for ChatGPT, said in a briefing after the keynote that the company “never meant to build a chatbot; we meant to build a super assistant, and we got a little sidetracked.”
Turley said that OpenAI is most excited about what it has achieved in natural language processing, but that the firm would continue to experiment with different user interfaces.
“Will people spend all of their time in ChatGPT? I don’t think so,” Turley said. “I can imagine you starting your day with ChatGPT,” then being guided toward other apps and websites.
At large, Dev Day 2025 appeared to be OpenAI’s declaration that it wants to own the next user interface of the internet. Apps that once lived in tabs may soon live in threads; clicks may give way to prompts; and “search” could evolve into “conversation.”
If OpenAI succeeds, ChatGPT could indeed become the new browser in a post-search era where consumers talk, transact and create inside one single conversational window.
However, as history shows, every major shift in how consumers use technology brings new questions — about openness, privacy, and who ultimately controls the platform. Those may just be getting started.