CourionAI
EN
Newsletter
← All news
google 2 min read

Android Studio Lets You Bring Your Own AI Coding Agent

At the Android Dev Summit, Google opened Android Studio to more than 40 third-party coding agents and added tools meant to make agents faster and cheaper to run.

Risograph illustration of a code editor with a row of small robot heads plugging into one coral socket like cartridges

Google used its Android Dev Summit to make a notable move: Android Studio, the main tool for building Android apps, no longer limits you to Google’s own AI. A feature called “Bring Your Own Agent” lets developers run more than 40 coding agents inside the editor, including Anthropic’s Claude Agent, OpenAI’s Codex and Google’s own Antigravity. They can be installed from a built-in marketplace and even run side by side, as The Decoder reports.

The plumbing behind this is the Agent Client Protocol, an open standard that lets an editor and an AI agent talk to each other, a bit like a USB port for coding assistants. Agents get access to useful parts of the project, such as its structure, the phone emulator and the error logs. Google also made its Android command line tool an official plugin for Claude Code, Codex and Antigravity. One trick lets agents preview an app’s screen in the terminal without starting an emulator. In Google’s own tests, agents finished interface tasks 77 percent faster and used up to three times fewer tokens (the pieces of text a model reads and writes, which you pay for). In another test with Claude Opus 5.5, a memory leak tool helped find and fix leaks nearly twice as fast with about 25 percent fewer tokens.

For app makers, Google also announced early building blocks for agents on the phone itself. The Agentic API in ML Kit uses Gemini Nano 4 to run small tasks on the device, and the AppFunctions API lets apps offer their features to assistants through the Model Context Protocol, a shared way for AI tools to call other software. Many of these are still alpha, beta or preview.

What’s behind it: Google is conceding that developers already have favourite agents and will not switch editors for them. By supporting all of them, it keeps Android Studio as the place where the work happens. It also gets Android knowledge, via more than 30 documentation packs called Android Skills, into every agent, which makes apps built with any of them more likely to follow Android’s rules.

A fair caveat: the speed and token numbers come from Google’s internal tests, and no outside group has checked them. Several of the new APIs are not final.

What this means for you: if you build Android apps, you can try your preferred agent without leaving Android Studio, which is the practical win. If you do not code, the signal is simply that coding assistants are becoming interchangeable parts rather than one locked product. Wearable fans may also note the announced XREAL AURA glasses at 1,099 dollars, shipping from mid November, though that is a separate story.

Sources

Source: https://the-decoder.com/google-opens-android-development-to-third-party-ai-agents-and-unveils-new-gemini-apis/

Next story

A Hidden Status Label in Claude Code Reportedly Eats Up to a Quarter of Usage

A developer's telemetry suggests Claude Code's background agents re-send their whole transcript every 30 seconds just to write a 3 to 5 word status line. Here is what was measured and what is still unverified.

Risograph illustration of a small label tag on a terminal window swallowing a coral slice of a fuel gauge while a magnifying glass looks on