OpenAI Is Still Building a "Super App." A Senior Employee Just Said "Chat Is Dead."
OpenAI is developing a super app to replace the chat interface entirely. Here's what that actually means for how you'll use AI in 2026 and beyond.
The chat box is going away. That's not speculation — it's the stated position of at least one senior OpenAI employee, who put it plainly: "Chat is dead."
OpenAI has been quietly building what it calls a "super app," and the work is still ongoing. The phrase gets thrown around in tech circles constantly, usually meaning little more than "we want you to spend more time in our product." But what OpenAI appears to be describing is something more structural: a shift away from the prompt-and-response model that has defined consumer AI since ChatGPT launched in late 2022.
This matters more than it sounds like it does.
What "Chat Is Dead" Actually Means
The chat paradigm — you type something, the AI replies, you type again — was always a stopgap. It was the simplest possible interface for getting large language model capabilities in front of users fast. It worked. Hundreds of millions of people adopted it. But it was never the end state.
What replaces it, according to what OpenAI has been building toward, is something closer to a persistent, proactive system that doesn't wait for you to ask. It knows your context. It initiates. It handles multi-step work across apps without you narrating every move.
That's the super app vision: less conversation, more delegation.
The timing here is notable. Every major AI platform is moving in the same direction right now. Microsoft shipped Copilot Wave 3 in March 2026, which brought autonomous multi-step task execution into Microsoft 365 via something called Copilot Cowork. Google's Gemini agentic platform got a major push at Cloud Next '26 in April. The "chat" wrapper is getting stripped away industry-wide, replaced by agent-first interfaces that act more like operating systems than assistants.
OpenAI is racing to make sure ChatGPT is that operating system for consumers before Microsoft owns it for enterprise.
Why the Super App Frame Is Significant
A super app isn't just a better chatbot. It's a platform play. Think of how WeChat works in China — payments, messaging, commerce, scheduling, all inside one container. OpenAI's version would presumably combine AI assistance, memory, tool use, voice, image generation, and task automation into a single product that people live inside.
That's a fundamentally different business than selling API access or monthly subscriptions to a chat interface. It's also a fundamentally different relationship with the user.
The memory angle is critical here. Current ChatGPT memory is limited and opt-in. A true super app would need persistent, rich memory to function — it needs to know who you are, what you're working on, and what you've already decided. That's a hard engineering and trust problem. If you've ever run into the frustrating limits of how AI tools handle continuity, the AI Memory Problem piece we ran earlier explains exactly why this is harder than it looks.
There's also the context problem. Proactive AI agents fail when they don't have enough context to act without asking for clarification every thirty seconds. OpenAI's super app will live or die on how well it solves that. It connects directly to something we've written about in depth: why AI tools don't know what you actually need even when you're talking to them constantly.
The Token Cost Problem Sitting Underneath All of This
Here's the tension nobody in the super app conversation is being honest about: running a persistent, proactive, memory-rich AI agent for hundreds of millions of users costs an extraordinary amount of money.
The industry is already struggling with this. Token costs are rising as companies push AI into more complex, multi-step tasks. Australia's Commonwealth Bank flagged surging AI costs publicly this month, noting costs will rise "in less predictable ways" as task complexity grows. That's a polite way of saying: the more useful this stuff gets, the more expensive it gets, and nobody has a clean pricing model for it yet.
OpenAI's current subscription tiers don't map cleanly onto a super app model. A $20/month ChatGPT Plus subscription made sense when the product was answering questions. It makes less sense if the product is autonomously managing your calendar, drafting your emails, running your research pipeline, and executing multi-app workflows while you sleep. That usage profile could consume orders of magnitude more compute per user.
The GitHub Copilot token billing fiasco is a preview of what happens when AI companies try to solve this problem mid-product. Developers were furious when Copilot switched to token-based pricing, because it turned a predictable expense into an unpredictable one. OpenAI will have to figure out a pricing model that makes a super app economically viable without alienating the users it's trying to attract.
What This Means for Competing Products
If OpenAI ships a credible super app, the products that feel this most immediately are the standalone AI tools that do one thing well. Meeting transcription tools, AI search layers, note-taking assistants — they all become features rather than products if a single app absorbs their use cases.
This is already happening at the edges. Lovable's bet on Google Cloud infrastructure reflects how even well-funded AI startups are thinking about their long-term positioning as the big platforms expand. The platform risk is real.
It's also worth watching what happens to OpenAI's own internal priorities. Greg Brockman returned to lead product strategy earlier this year, and the super app project almost certainly falls within his remit. Whether the product ships with the coherence the vision requires is a different question from whether OpenAI is serious about building it — and they clearly are.
What You Should Actually Do
If you're a professional who depends on AI tools daily, the super app shift changes your calculus in one specific way: it's worth being deliberate now about which platform you're building your workflows around, because switching costs will go up as these products get stickier.
A few things worth doing before this landscape consolidates:
- Audit your current AI stack. Which tools are you actually using daily versus tools you subscribed to and forgot? The super app era will reward consolidation.
- Pay attention to memory and continuity features. Any AI tool that's improving how it retains context across sessions is moving in the direction the market is heading. Any tool that still resets every conversation is moving away from it.
- Don't over-index on any single platform yet. OpenAI hasn't shipped the super app. Microsoft's Copilot is compelling for enterprise but limited for consumers. Google's Gemini is accelerating. The winner isn't decided.
- Watch the pricing changes. If OpenAI announces a new pricing tier in the next six months that looks consumption-based rather than flat, that's the tell. It means the super app is closer than the announcements will let on.
The chat interface built the AI industry as we know it. The companies that replace it will define the next phase. OpenAI is betting it gets to do both.
