How AI Could Render the App Obsolete as an Interface

Over the past decade and a half, the software industry trained users to download specialized applications for nearly every task: email, music, banking, photo editing, food delivery, flight booking, spreadsheets, note-taking and even app discovery itself. But artificial intelligence could render this fragmented approach obsolete—not by destroying software companies overnight, but by fundamentally changing how people interact with digital tools. Rather than learning interfaces built around a company's product, users might soon communicate intentions to a single intelligent intermediary that orchestrates dozens of services behind the scenes.

This transition is already underway. Apple has made app actions and data accessible to Apple Intelligence outside individual applications through App Intents. OpenAI has integrated external apps directly into ChatGPT, while Google's Gemini can interact with connected applications to complete tasks. The infrastructure for a post-app world is being constructed quietly while the industry continues debating interface minutiae.

The App Was Always a Compromise

Consider the current workflow for accomplishing a goal. A user decides what they need, identifies which application can deliver it, launches that application, navigates its interface, locates the relevant feature and enters information in the format the software demands. Often this process repeats across multiple applications.

Planning a vacation illustrates this friction: bouncing between a web browser, mapping tools, airline websites, hotel applications, calendar software, email, note-taking apps, weather services and possibly spreadsheets. The human effectively becomes the API connecting disparate systems.

Artificial intelligence inverts this dynamic. Instead of opening eight separate tools, a user could simply state: "Plan a five-day trip to Tokyo in April. Keep the hotel under $250 a night, avoid overnight flights, put everything on my calendar and make me a packing list based on the forecast." This represents something fundamentally different from traditional app interaction—it is an intention, and intention may emerge as computing's most critical interface.

Nobody Actually Wants to Use Software

The software industry rarely acknowledges an uncomfortable reality: most people do not want software. They want outcomes. Nobody celebrates opening a spreadsheet; they want financial clarity. Nobody yearns to navigate airline booking interfaces; they want to reach their destination. Nobody aspires to master photo editor menus; they want unwanted people removed from images.

For decades, software interfaces served as necessary intermediaries between human intention and computational execution. As AI begins eliminating portions of that intermediary layer, it represents a shift potentially more significant than adding a chatbot to existing products.

ChatGPT demonstrates an early version of this model. OpenAI's app system allows services to appear directly within conversations, enabling users to interact with external services without launching standalone applications. Google pursues a comparable strategy through Gemini's Connected Apps feature, which can leverage connected services to execute tasks including calendar event creation, message sending, music playback and interactions with other supported applications. The app persists technically, but the user's relationship with it transforms.

Your Phone Could Eventually Have One App

Envision unlocking a phone to find minimal visual elements: a single input field. A user types, speaks, photographs something or points their camera and requests: "Pay my electricity bill." "Find the photo Sarah sent me last summer." "Move my Thursday meeting because my flight changed." "Turn these vacation clips into a 30-second video." "Cancel subscriptions I haven't used in six months." Or simply: "Order what I normally get from the supermarket, except we don't need coffee."

Behind the interface, dozens of services continue operating. Banks process payments. Calendars store appointments. Supermarkets fulfill orders. Video engines render content. Yet users need not interact directly with any of them. The AI handles the orchestration. Applications do not vanish technically; they vanish psychologically, becoming infrastructure.

Apple's developer tools offer insight into this possibility. The App Intents framework specifically exposes app actions and data outside the application itself, including to Apple Intelligence, Siri, Spotlight and Shortcuts. Apple describes App Intents as enabling actions and content to be available to "the rest of the system." This represents a significant shift: the app ceases to be necessarily the destination, with its capabilities summoned from elsewhere instead.

This Could Be Terrible News for UI Designers

The design industry faces an uncomfortable reckoning. Interface design has operated under a foundational assumption for decades: people require interfaces. This drove the creation of dashboards, menus, buttons, forms, navigation systems, onboarding flows, settings screens, filters, dropdowns, sidebars and countless hamburger menus. But what occurs when users stop navigating?

If someone can simply instruct a computer, "Show me which clients haven't paid invoices older than 30 days and draft friendly reminders," why should they learn where Accounts Receivable resides in a dashboard? Substantial portions of user interface exist primarily because software previously lacked the ability to comprehend user intent. Increasingly, AI possesses this capability.

Designers will not disappear, but the subject of design could transform substantially. Rather than mapping every possible path through an application, designers may increasingly specify intentions, permissions, confirmations, exceptions, trust signals and moments requiring human intervention. Interface design shifts from navigation toward negotiation.

The designer's role may actually become more critical. When an AI prepares to transfer $5,000, cancel a flight or delete a decade of photographs, someone must design precisely how the system communicates its pending action, obtains permission and enables error correction. The button may vanish. The responsibility persists.

Websites Could Be Next

Applications are not the sole endangered interface. Websites face identical pressures. Why browse 300 hotels when an AI understands your preference for quiet boutique properties, dislikes carpeting, wants walkable neighborhoods, requires a gym and refuses $40 breakfast charges? Why scan restaurant listings when your assistant already knows you seek something within a 15-minute walk, moderately priced, open tonight and unlikely to demand shouting across tables?

Why visit ten product pages when an AI can potentially compare specifications, reviews, return policies and pricing? For numerous websites, the uncomfortable prospect extends beyond simply receiving less traffic from AI. The possibility emerges that users eventually need not visit them at all. The website—or more precisely, the services and data underlying it—becomes something machines query and negotiate with. For web designers, this represents a more consequential threat than "AI can create websites now." The genuine disruption may be that fewer interactions require a traditional website.

The App Store Could Become an AI Tool Store

An alternative scenario exists: applications persist, but humans cease selecting and launching them as they currently do. AI agents increasingly determine which service or capability to invoke based on user objectives. Manipulating a PDF? The assistant activates a PDF tool. Booking a hotel? It connects to a reservation service. Creating a chart? It calls a visualization tool. Arranging delivery? It connects to a courier. Users would not necessarily care which software was employed any more than they care which network service loaded this webpage.

Model Context Protocol, or MCP, originally introduced by Anthropic as an open standard for connecting AI systems with external tools and data, makes this scenario plausible. Anthropic's connectors now permit Claude to access connected tools and, where authorized, execute actions within them. OpenAI's Apps SDK similarly builds on MCP to connect ChatGPT with external tools and services.

This shift carries competitive implications. Winning software companies in this environment may not necessarily possess the most polished interfaces. They might prevail because they offer superior data, APIs, reliability, permissions and agent integrations. For years, companies competed to become the application users opened. Tomorrow they may compete to become the tool an AI selects.

And Yes, There's a Giant Problem

Granting one AI system access to an entire digital life sounds wonderfully convenient until examined closely. Who controls it? Who accesses the data? Who determines which hotel receives recommendations? Can companies pay to become the preferred option? What happens when the system misinterprets "cancel my meeting" and cancels a flight instead? What happens when it spends $3,000 instead of $300?

What happens when the company controlling the universal interface raises prices, injects advertising or prioritizes its own services? Today's fragmented app ecosystem proves annoying, but it also represents a form of decentralization. Replacing 100 interfaces with one extraordinarily powerful intermediary introduces an entirely new risk category. Suddenly the company positioned between users and the broader digital world possesses extraordinary power over what users see, which services they access and potentially what they purchase.

This explains why traditional interfaces probably will not completely vanish. Sometimes users need to see the spreadsheet. Sometimes they want to browse every hotel themselves. Sometimes a prominent confirmation button represents exactly what good design demands. AI may eliminate substantial interface friction without eliminating interfaces entirely.

The Irony: Apps May Have to Become Invisible to Survive

Technology companies fought for years to place their icon on home screens. The next battle could reverse this: they will fight to become the invisible service an AI selects. The superior weather company will not necessarily be the one whose app users open. It will be the one an assistant trusts when asked whether an umbrella is necessary. The superior travel company might be the one an AI quietly contacts when someone says, "Get me to Barcelona." The superior design software might be the engine operating behind the instruction, "Make this presentation look expensive."

Brands will not disappear. Software will not disappear. APIs will not disappear. Databases certainly will not disappear. But the small rectangular window through which companies force interaction with all of that? That suddenly appears vulnerable. Perhaps this represents the ultimate irony of the AI revolution: software could become more powerful and more important than ever while simultaneously becoming less visible to the people using it.

We May Be Designing the Last Generation of Apps

Every major computing era has eliminated a layer of friction. Command lines gave way to graphical interfaces. Desktop software gave way to websites. Websites gave way to mobile applications. Each transition did not completely destroy what preceded it; it pushed additional complexity underneath a simpler layer. AI could represent the next layer.

This time, however, it challenges something more fundamental: the assumption that people should have to operate software at all. Perhaps the ultimate interface is not a superior dashboard. Perhaps it is no dashboard—or at least no dashboard until one is actually needed. Users explain what they want. The machine determines which tools are required, requests permission when necessary, completes the work and displays the result. The applications remain, the companies remain and the software continues running. Users simply stop thinking about most of it.

If this occurs, the greatest threat AI presents to the app industry is not that it will construct applications faster. It is that eventually, people may stop caring whether an app exists at all.