4 Ways to Prepare for the Rise of Disposable User Interfaces
▼ Summary
– Salesforce’s “Headless 360” exposes its platforms as APIs, allowing agents to access data without a browser UI, signaling a shift away from traditional interfaces.
– Michael Grinich predicts the “UI era” is ending, with interfaces becoming disposable, generated on-demand and context-aware via AI, rather than fixed screens.
– UIs are shifting from user-operated tools to “systems that produce outcomes,” where users express intent and AI creates the interface and actions.
– The user role evolves from operator to collaborator and director of AI agents, requiring a change in how technology professionals approach development.
– Grinich advises that the product is now the capability and data, not the UI; APIs are the real surface, and the model itself becomes the interface, reducing cognitive overload.
The era of the traditional user interface is coming to an end. With Salesforce’s recent launch of Headless 360, the company’s platforms,Salesforce, Agentforce, and Slack,are now accessible through APIs, MCP, and CLI, allowing AI agents to pull data, execute workflows, and complete tasks without any browser-based UI. This shift signals a broader transformation: the future of software is no longer about designing beautiful, clickable screens for humans, but about building systems that serve autonomous agents.
At the TypeScript AI Demo Day in San Francisco, WorkOS founder Michael Grinich declared, “We are exiting the UI era.” He described a world where disposable interfaces are generated on demand, appearing as simple text boxes or one-time-use projections that vanish after a single interaction. “When you need a new one, just make a new interface,” he explained. In this model, users no longer operate static screens; they express intent, and AI models generate the UI and actions in real time.
This evolution makes computing more human-centric, even as it becomes more automated. Generative AI, one of the fastest-growing technologies ever, presents a simple text box asking, “What do you want?” Grinich traced UI history from switches to commands, pointers, cursors, touch, and now to language. “Due to language models, we’ve had this breakthrough where the UIs are now synthesized per request, just in time for you. They’re context aware.”
For technology professionals, this transition demands a new mindset. Grinich offered four key strategies:
First, UI is no longer the product. The real value lies in the capability, model, and data that come together. “The UI is actually just a projection layer of all that,” he said.
Second, the components still matter, but not in the traditional sense. “You’re giving elements to the model, and the model is figuring out what to do with it,” Grinich noted. Developers must provide the right building blocks for large language models to assemble context-appropriate interfaces.
Third, APIs become the real surface you build on. Agents don’t click buttons; they consume APIs. “The UI is no longer a product,it’s the API,” Grinich emphasized.
Fourth, the model itself is the interface. As interfaces shrink to APIs and data layers, the goal is reducing cognitive overload. Grinich compared this to modern cars that minimize physical controls in favor of digital, autonomous systems. “You don’t really care about driving. You care about getting to your destination.”
Y Combinator’s classic advice is to “make something people want.” Grinich proposed an update: “Make something that agents want. The agents will be doing things for people. If you want to serve people, you need to serve their agents, too.”
(Source: ZDNet)
