Google has released a new feature called Agent Mode for developers in Android Studio, an experimental capability that uses its flagship artificial intelligence model Gemini to handle complex developmental tasks. This new mode integrates agentic AI capabilities, allowing developers to spell out complex goals using their own words and receive guidance from the agent.
With Agent Mode, developers can generate unit tests, perform intricate code refactoring, and execute plans under their guidance. The feature employs various tools in the editor for reading and modifying code, working on projects, searching the codebase, and managing lengthy tasks with minimal oversight.
While powerful, developers remain in control of Agent Mode, with the ability to refine and guide agent outputs and choose to accept or reject them. Users can also enable “auto-approve” if they don’t want to be bothered.
A paid subscription to Gemini is required to expand the context window using Gemini 2.5 Pro’s 1 million token context window, which can span enormous codebases. This larger context window provides higher-quality responses and allows developers to send more instructions and attachments to Gemini for better results.
The latest experimental preview version of Android Studio with Agent Mode can be downloaded on the canary release channel of Narwhal today for users to try out.
Source: https://siliconangle.com/2025/06/24/google-releases-ai-powered-agent-mode-developers-android-studio