Google Just Killed the App

How Gemini Intelligence turns Android from an operating system into an AI agent, and why a $1 trillion app economy is about to find out what happens when users stop opening apps

Author

Justin Wright
May 18, 2026 • Est. Reading Time: 5 minutes

Happy Monday!

At the Android Show on Monday, Google's head of Android Sameer Samat said something that should have gotten more attention than it did: "We're transitioning from an operating system to an intelligence system."

That is not just marketing language. It is a description of what Gemini Intelligence actually does. Starting this summer on flagship Samsung and Pixel phones, Gemini will be able to read what is on your screen, move across apps, and complete multi-step tasks autonomously. Long-press a grocery list in your notes and Gemini builds a shopping cart across retailers. Need a ride, dinner reservation, and a price comparison? Gemini handles all three without you opening a single app.

The feature set is impressive, but the implication is enormous. The entire mobile app economy, a market projected to exceed $1 trillion this year, is built on a single assumption: users open apps. If an AI agent handles tasks across apps on your behalf, that assumption breaks. And when it breaks, so does the business model underneath it.

Google announced Gemini Intelligence at I/O, turning Android from an operating system into an AI agent that moves across apps and completes tasks autonomously. eMarketer estimates this could disrupt more than two-thirds of the mobile app economy. Google also unveiled Googlebooks, laptops designed from scratch around Gemini, and is rebuilding Android Auto for 250 million cars. Apple is rushing to match with a Gemini-powered Siri reboot at WWDC next month. The companies that built the app store are now betting that the intelligence layer above it is worth more than the store itself.

TL;DR

What Gemini Intelligence Actually Does

The core capability is cross-app automation. Instead of opening individual apps to accomplish tasks, Gemini operates as a layer above them, reading context from your screen, navigating between services, and executing multi-step workflows. Google demonstrated scenarios where Gemini handles ride-hailing, food orders, and form submissions without the user ever launching an app directly.

The system runs only in apps you permit, will not automate tasks without instruction, and gives users granular controls over what data Gemini can access. Google is clearly aware of the privacy implications: Gemini Intelligence needs access to emails, Drive files, calendar entries, location data, and payment methods. That concentration of personal data inside a single AI agent is a legitimate concern, but Google is betting that the convenience trade-off wins.

The hardware requirements are notable: a flagship chip, 12GB of RAM, and specific on-device AI models. This is not coming to budget phones. It rolls out on Samsung Galaxy and Pixel flagships this summer, then expands to watches, cars, glasses, and laptops later this year.

The $1 Trillion Question

eMarketer published a briefing the same week estimating that Gemini Intelligence could disrupt more than two-thirds of the mobile app economy. The logic is straightforward: when Gemini handles ride-hailing, food orders, and price comparisons autonomously, users no longer need to open individual apps. They do not see product ads, encounter loyalty prompts, or browse. The entire discovery and engagement layer that funds mobile software becomes invisible.

The app economy generates over $1 trillion annually through a combination of in-app purchases, subscriptions, and advertising. All three revenue streams depend on users actively opening and engaging with apps. An AI agent that completes tasks across apps on the user's behalf short-circuits that engagement loop entirely.

App downloads are still growing, but downloads without engagement are meaningless. If Gemini handles the task, the app is installed but never opened. The metric that matters shifts from downloads to whether your app is compatible with Gemini's automation layer.

Googlebooks and the Hardware Play

Google also unveiled Googlebooks, laptops designed from the ground up around Gemini Intelligence. Partners include Acer, ASUS, Dell, HP, and Lenovo, launching this fall. The signature feature is a glowing light strip on the lid that activates when Gemini is running. Google wants AI to be visually present and fundamentally built into the overall experience.

The broader play is cross-device continuity. Gemini Intelligence rolls out across phones, laptops, watches, cars (Android Auto serves 250 million vehicles), and eventually XR glasses. Google is building an intelligence layer that follows you across every screen. The operating system becomes secondary to the AI that runs on top of it.

Apple Is a Month Behind

Google timed this announcement intentionally. WWDC is June 8, less than a month away. Apple is expected to unveil a rebuilt Siri powered by Google's own Gemini models, the result of a partnership announced in January worth an estimated $1 billion annually.

The rebuilt Siri will function as a full chatbot with persistent history, cross-app awareness, and multi-step command execution. It is the same vision Google just demonstrated, but Apple has been promising it since iOS 18 in 2024. It is now expected in iOS 27 this fall.

Apple's answer to Google's intelligence layer is built on Google's model. Both platforms are converging on AI agents that work across apps, but Google shipped first. For developers deciding where to invest, that head start matters.

What This Means for Practitioners

For mobile developers, this is an existential question. If Gemini handles tasks across apps, do you build for the intelligence layer or keep building traditional app experiences? The brands that move first to become default Gemini automation partners for categories like ride-sharing, reservations, and food delivery gain a structural advantage. Everyone else risks becoming invisible to users who never open the app directly.

For product leaders, the engagement metrics you track today may be meaningless within a year. Session duration, daily active users, and screens per visit all assume the user is inside your app. When an AI agent completes the task on their behalf, those metrics go to zero even if your app is technically being used. The new metric is Gemini compatibility: whether your service is discoverable and executable through the intelligence layer.

For anyone building on mobile distribution, Google is cannibalizing its own $60 billion app store to control the intelligence layer above it. That is the same logic that drove Google to make Android free in the first place: give away the operating system to control search. Now it is giving away app access to control the AI agent. Google always bets on the layer above.

The Bottom Line

Google just declared that the operating system is over; the intelligence system is what matters now. Gemini reads your screen, moves across your apps, and completes tasks you used to handle yourself.

The $1 trillion app economy was built on the assumption that users open apps. Google is betting that assumption is about to break, and that controlling the AI layer above the apps is worth more than the app store revenue it will cannibalize.

The companies that built the app store are now building the thing that makes it obsolete. If you are a developer, a brand, or a business that depends on mobile engagement, the question is no longer whether AI agents will change your distribution model. It is whether you will be visible when they do.

In motion,
Justin Wright

If an AI agent can handle your ride, your dinner, and your shopping without ever opening an app, what happens to the trillion-dollar economy built on the assumption that opening the app is necessary for these experiences?

Food for Thought

If you haven’t listened to my podcast Mostly Humans: An AI and business podcast for everyone yet, new episodes drop every week!

Episodes can be found below - please like, subscribe, and comment!