OpenAI is consolidating its scattered tools into one desktop app, and the real story is what they're killing to do it.
The Summary
- OpenAI is merging ChatGPT, Codex (coding agent), and Atlas (browser) into a single desktop "superapp", according to someone familiar with the plan
- CEO of Applications Fidji Simo told staff to deprioritize "side projects" that don't feed the core product
- This is OpenAI betting on integration over experimentation, one interface over many
The Signal
OpenAI spent years spinning up new products: ChatGPT for conversation, Codex for developers, Atlas for browsing. Each had its own interface, its own learning curve, its own reason to exist. Now they're collapsing that constellation into one thing. The message is clear: we're done exploring. We're building the control center.
This matters because it changes how people will interact with AI agents. Right now, if you want an agent to write code, you open one tool. If you want it to research something on the web, you open another. If you want to chat, you're back in ChatGPT. That friction is a feature for OpenAI's competitors. Every context switch is a chance to lose you to Claude, Gemini, or whoever builds a better single experience. The superapp play eliminates that. One window, one login, one place where your agent does everything.
Fidji Simo's directive to kill side projects is equally telling. She came from Meta, where she ran the Facebook app, the ultimate digital black hole that tried to keep you from ever leaving. That DNA is showing. OpenAI isn't just consolidating tools. They're consolidating *attention*. The goal is to make the superapp the last app you need open. Your agent writes the code, tests it in the browser, debugs it in the chat interface, all without you tabbing away.
This is the Agent Economy endgame: not better tools, but fewer tools. Not more specialized agents, but one generalist agent that knows how to route your request to the right capability. The companies that win Web4 won't be the ones with the most features. They'll be the ones that make orchestration invisible.
The Implication
If you're building agent tooling, watch what gets left out of OpenAI's superapp. Those gaps are your opportunity. The "side projects" getting killed are someone else's main product. For users, start thinking about what you actually need from multiple tools versus what you could do in one if it were smart enough. The superapp era is coming, and it won't just be OpenAI. Every major AI lab is probably sketching the same diagram right now.
Source: The Information