Google is opening its Android Studio development environment to third-party AI agents and unveiling new Gemini APIs during the Android Dev Summit. The update introduces the "Bring Your Own Agent" feature, allowing developers to utilize over 40 external coding models such as Anthropic's Claude and OpenAI's Codex directly inside the IDE.
To support these integrations, Google launched version 2.0 of the Android Bench leaderboard, alongside new terminal tools and specialized skills designed to reduce token usage and speed up development tasks. Developers can now run agent functions locally on mobile devices using the Gemini Nano 4-powered Agentic API and the Model Context Protocol. Furthermore, Google and hardware maker XREAL revealed the XREAL AURA, a pair of smart glasses running Android XR that uses Gemini to analyze real-world environments.
This release matters because it fundamentally shifts mobile software development toward multi-model AI agent workflows, lowering friction for third-party tools within Google's core ecosystem.