OpenAI Is Merging ChatGPT, Codex, and Atlas Into One Desktop Superapp — Here's What Changes for Your Team
OpenAI is unifying ChatGPT, Codex, and its Atlas browser into a single desktop superapp with Canva and Booking.com integrations built in.
OpenAI is done running three separate products and hoping users figure out how they fit together. The company is merging ChatGPT, its Codex coding assistant, and its Atlas browser into a single desktop application, with Canva and Booking.com integrated directly at launch. The project is internally codenamed “Aria,” and according to reporting from the Financial Times citing more than a dozen current and former employees, it is now in active rollout.
This is not a cosmetic redesign. It is a structural decision about what OpenAI wants to be.
What is actually changing
Until now, OpenAI’s products lived in separate places. ChatGPT handled conversation and general tasks. Codex was a standalone coding environment aimed at developers. Atlas, OpenAI’s AI-native browser, sat to the side. If you used all three, you were context-switching constantly.
The superapp puts all of that under one roof on the desktop. Codex becomes an agent you can invoke from anywhere inside the application, not just a dedicated coding tab. Atlas provides the browser layer, so you can interact with ChatGPT in the context of whatever you are looking at, rather than copying content across windows. Partner integrations from Canva and Booking.com are built in at launch, with Expedia, Figma, Spotify, Coursera, and Zillow also in the pilot.
The payments layer runs on a commerce protocol OpenAI co-developed with Stripe, meaning the assistant can complete purchases, not just recommend them.
Fidji Simo, OpenAI’s CEO of Applications, is leading the effort. Her internal memo was direct: “Fragmentation has been slowing us down and making it harder to hit the quality bar we want.” OpenAI President Greg Brockman is temporarily overseeing the product and organisational changes that come with it.
One senior employee told the Financial Times that “chat is dead.” What they meant was that the pure question-and-answer model is being replaced by a platform that actually does things.
The commercial logic behind it
This is not purely a product decision. OpenAI filed a confidential IPO draft in May 2026, targeting a debut as soon as September at a valuation somewhere between $730 billion and $850 billion. How you get from “chatbot company” to a valuation in that range is by demonstrating you own a platform, not just a popular app.
The numbers support the direction. Codex grew its weekly active user base sixfold to more than 5 million users since launch. OpenAI’s 2 million business customers currently account for roughly 40% of revenue, and the company is targeting 50% by the end of the year. A unified product sold as a single enterprise subscription is a cleaner commercial story than three separate tools with different pricing tiers and different procurement conversations.
There is also a competitive dimension. Anthropic’s Claude Code hit $2.5 billion in annualised revenue by February 2026 and has been winning developer mindshare. OpenAI’s answer is not to out-feature Claude Code in isolation, but to make Codex part of a broader platform that no standalone coding assistant can match.
What this means for teams running ChatGPT and Codex today
If your team currently uses ChatGPT for knowledge work and Codex separately for development tasks, the practical change is consolidation rather than disruption. The workflows you have built do not break. What changes is where they live and how they connect.
Codex running inside the superapp can be invoked by other parts of the system, which means a workflow that previously required switching between tools can eventually be handled in sequence within one interface. Simo has described wanting Codex to power general-purpose task completion across all of OpenAI’s products, not just programming.
The mobile side of this is already live in a limited form. OpenAI launched Codex integration in the ChatGPT mobile app in May, letting developers monitor and manage coding tasks from their phones while the actual environment runs on a connected desktop. Files and credentials stay off the device. The superapp itself, though, is specifically a desktop product. The mobile app is not changing as part of this consolidation.
For enterprise buyers, the practical impact is that one procurement conversation now covers use cases that previously required three. Pricing for the unified platform has not been disclosed, and reconciling the existing tier structures (Codex and Atlas agent mode both require Plus or above, API access is pay-per-use) will be one of the more consequential decisions OpenAI makes before general availability.
A note on the integrations if you are in the UK or EEA
The Canva and Booking.com integrations are currently unavailable in the UK, the European Economic Area, and Switzerland. OpenAI lists these regions as “coming soon” with no committed date. UK and EEA users will get the interface redesign, but the third-party integrations that define the superapp experience could be months away.
The technical layer worth knowing about
The integrations run on the Model Context Protocol (MCP), an open standard that lets the model discover and call external tools during a conversation. It was originally developed by Anthropic and has since become the de facto standard for this kind of integration work.
MCP is powerful but not without risk. A vulnerability tracked as CVE-2025-6514, rated 9.6 out of 10 on the severity scale, showed that a malicious server could smuggle instructions through a tool description. An assistant that can read your calendar, hold a payment token, and act across a dozen services concentrates exactly the kind of access attackers look for. If you are evaluating this platform for enterprise use, MCP security posture should be on your checklist.
The broader direction
The superapp model, popularised by WeChat and Grab in Asia, means a single platform handles chat, productivity, payments, and commerce. OpenAI is building the AI equivalent, targeting the professional desktop as its starting point.
The long-term goal, according to people familiar with the project, is eventually to remove the need for prompts at all. The system would learn your habits and act without being asked. That is some distance away. What is arriving now is a consolidated desktop platform that makes OpenAI’s products easier to buy, easier to use together, and easier to build a commercial case around.
For teams already using ChatGPT and Codex, the near-term message is: your tools are moving under one roof. The workflows stay. The interface gets simpler. And the pricing clarity, when it arrives, will matter.