Google Unveils AppFunctions to Connect AI Agents and Android Apps

In a significant shift towards an “agent-first” operating system, Google has unveiled early beta features aimed at enhancing Android’s task-centric capabilities. This initiative allows applications to serve as functional building blocks, which users can leverage through AI agents or assistants to achieve their objectives.

Introducing AppFunctions

The cornerstone of this innovative model is AppFunctions, a Jetpack API designed for developers to expose self-describing capabilities within their applications. This integration enables seamless interactions with AI agents while prioritizing user privacy and performance by executing tasks on-device, thereby reducing network latency.

Mirroring how backend capabilities are declared via MCP cloud servers, AppFunctions provides an on-device solution for Android apps. Much like WebMCP, it executes these functions locally on the device rather than on a server.

Consider a scenario where a user requests the Gemini Assistant to “Show me pictures of my cat from Samsung Gallery.” The assistant interprets the request, retrieves the relevant images, and displays them within its interface. This contextual persistence allows users to reference these images in subsequent requests, whether for editing, sharing, or taking further actions.

UI Automation Platform

Recognizing that not all applications will initially support AppFunctions, Google has also introduced a UI automation platform within Android. This layer serves as a fallback, enabling users to perform complex tasks effortlessly. For instance, they can place intricate pizza orders for family members with varying preferences, coordinate multi-stop rideshares with colleagues, or reorder groceries—all through the Gemini Assistant without necessitating additional developer input.

This is the platform doing the heavy lifting, so developers can get agentic reach with zero code. It’s a low-effort way to extend their reach without a major engineering lift right now.

In its announcement, Google underscored that privacy and user control are paramount in the design of AppFunctions. All interactions are crafted for on-device execution, ensuring full user visibility through live views and notifications. Users retain the ability to manually override the agent’s actions, with mandatory confirmations required for sensitive tasks such as purchases.

Currently, AppFunctions and the UI automation platform are in early beta, available exclusively on the Galaxy S26 series, with plans for broader deployment in Android 17.

About the Author

Sergio De Simone


Show moreShow less

AppWizard
Google Unveils AppFunctions to Connect AI Agents and Android Apps