⌘K
Change language Switch ThemeSign In
Narrow Mode
Google Unveils AppFunctions to Connect AI Agents and Android Apps
I InfoQ @Sergio De Simone
One Sentence Summary
Google introduces AppFunctions, a Jetpack API enabling Android apps to expose capabilities to AI agents for on-device task execution, alongside a UI automation fallback.
Summary
Google is shifting Android toward an 「agent-first」 OS with the introduction of AppFunctions and a UI automation platform. AppFunctions is a Jetpack API that allows developers to expose app capabilities as self-describing modules for AI agents like Gemini to invoke locally. This on-device approach prioritizes privacy and speed by minimizing network latency. For apps without native integration, a new UI automation layer acts as a fallback, allowing agents to perform complex tasks—such as ordering food or coordinating rideshares—by interacting with existing app UIs without additional developer effort. Currently in beta on the Galaxy S26, these features are slated for a wider rollout in Android 17.
Main Points
* 1. Introduction of AppFunctions for agent-app synergy.A new Jetpack API that lets Android apps expose functional building blocks to AI agents, allowing them to perform specific tasks within apps seamlessly and programmatically. * 2. On-device execution for privacy and performance.Unlike cloud-based solutions, AppFunctions runs locally on the device, which reduces network latency and ensures that sensitive user interactions remain private. * 3. UI automation as a zero-code fallback mechanism.For apps that have not yet integrated the AppFunctions API, Android provides a platform-level automation layer that allows agents to navigate app UIs to complete complex user requests. * 4. Focus on user control and transparency.The architecture includes mandatory confirmations for sensitive actions and live visibility of agent activities through notifications to ensure the user remains in command.
Metadata
AI Score
91
Website infoq.com
Published At Today
Length 364 words (about 2 min)
Sign in to use highlight and note-taking features for a better reading experience. Sign in now
In a move to transform Android into an "agent-first" OS, Google has introduced new early beta features to support a task-centric model in which apps provide functional building blocks users leverage through AI agents or assistants to fulfill their goals.
The foundation for this new model is provided by AppFunctions, a Jetpack API that allows developers to expose self-describing capabilities within their apps for seamless integration with AI agents. By running on-device, these interactions offer improved privacy and faster performance by minimizing 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.
For example, a user might ask Gemini Assistant to "Show me pictures of my cat from Samsung Gallery". The assistant would interpret the user's request, retrieve the relevant photos, and present them in its own interface. Those images can then persist in context, allowing the user to reference them in follow-up requests, such as editing, sharing, or taking further action.
As not all apps will support AppFunctions, especially in this early stages, Google has also introduced an UI automation platform in Android that provides a fallback when apps aren't integrated. This automation layer makes it possible for users to "place a complex pizza order for their family members with particular tastes, coordinate a multi-stop rideshare with co-workers, or reorder their last grocery purchase" all through the Gemini Assistant without additional developer effort.
> 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 emphasized that privacy and user control are central to the design of AppFunctions. All interactions are built for on-device execution with full user visibility through live view and/or notifications, the ability to manually override the agent's behavior, and mandatory confirmation required for sensitive actions such as purchases.
As noted, AppFunctions and the UI automation platform are still in early beta, currently available on the Galaxy S26 series, with a wider rollout of these features planned for Android 17.
I InfoQ @Sergio De Simone
One Sentence Summary
Google introduces AppFunctions, a Jetpack API enabling Android apps to expose capabilities to AI agents for on-device task execution, alongside a UI automation fallback.
Summary
Google is shifting Android toward an 「agent-first」 OS with the introduction of AppFunctions and a UI automation platform. AppFunctions is a Jetpack API that allows developers to expose app capabilities as self-describing modules for AI agents like Gemini to invoke locally. This on-device approach prioritizes privacy and speed by minimizing network latency. For apps without native integration, a new UI automation layer acts as a fallback, allowing agents to perform complex tasks—such as ordering food or coordinating rideshares—by interacting with existing app UIs without additional developer effort. Currently in beta on the Galaxy S26, these features are slated for a wider rollout in Android 17.
Main Points
* 1. Introduction of AppFunctions for agent-app synergy.
A new Jetpack API that lets Android apps expose functional building blocks to AI agents, allowing them to perform specific tasks within apps seamlessly and programmatically.
* 2. On-device execution for privacy and performance.
Unlike cloud-based solutions, AppFunctions runs locally on the device, which reduces network latency and ensures that sensitive user interactions remain private.
* 3. UI automation as a zero-code fallback mechanism.
For apps that have not yet integrated the AppFunctions API, Android provides a platform-level automation layer that allows agents to navigate app UIs to complete complex user requests.
* 4. Focus on user control and transparency.
The architecture includes mandatory confirmations for sensitive actions and live visibility of agent activities through notifications to ensure the user remains in command.
Key Quotes
* In a move to transform Android into an 「agent-first」 OS, Google has introduced new early beta features to support a task-centric model. * AppFunctions, a Jetpack API that allows developers to expose self-describing capabilities within their apps for seamless integration with AI agents. * This automation layer makes it possible for users to 「place a complex pizza order... coordinate a multi-stop rideshare... or reorder their last grocery purchase」 all through the Gemini Assistant without additional developer effort. * All interactions are built for on-device execution with full user visibility through live view and/or notifications.
AI Score
91
Website infoq.com
Published At Today
Length 364 words (about 2 min)
Tags
Android
AppFunctions
AI Agents
Gemini
Jetpack API
Related Articles
* Engineering Speed at Scale — Architectural Lessons from Sub-100-ms APIs * OpenAI Introduces Harness Engineering: Codex Agents Power Large‑Scale Software Development * 4 Patterns of AI Native Development * Architecting Agentic MLOps: A Layered Protocol Strategy with A2A and MCP protocol for inter-agent communication with the Model Context Protocol (MCP) for standardized tool and data access.") * Major Move: OpenClaw Creator Joins OpenAI * Harness design for long-running application development for long-running autonomous coding, detailing how to overcome context anxiety and self-evaluation bias to build complex full-stack applications.") * Andrej Karpathy on Code Agents， AutoResearch， and the Loopy Era of AI * The Ideal Micro-Frontends Platform * Anthropic Releases Claude Sonnet 4.6 with 1M Context Window * Anthropic Introduces Claude Opus 4.6 with 1M Token Context HomeArticlesPodcastsVideosTweets