AI coding assistants can struggle with complex build scripts. Deeply nested Gradle configurations, custom plugins, and heavy sync cycles often trip up automated tools and slow down developer machines. At the same time, apps need to run well outside standard phone displays, handling input from autonomous agents, connected devices, and spatial headsets.

This session looks at how the Android stack is adapting to these workflows. We explore how Lightbuild and the Android CLI introduce a declarative layer that works alongside existing Groovy and Kotlin DSL setups, giving AI agents predictable schemas they can inspect and edit safely. From there, we move to runtime architecture: turning app logic into callable tools using Android MCP and Jetpack AppFunctions, rendering dynamic agent output directly in Compose with A2UI, and bringing those experiences into 3D using the Jetpack XR SDK.

The talk features working sample repositories and real architecture patterns, not slideware or product hype. Demos cover experimental and early-access platform tools (Lightbuild, Jetpack A2UI, Android MCP, and Jetpack XR SDK); expect rapid API iterations as these libraries move toward stable release. Basic familiarity with modern Android architecture, Jetpack Compose, and build workflows is recommended.