Google is entering the growing market for AI-assisted programming with a new tool called "Jules." The agent is designed to help developers with repetitive tasks like fixing bugs, writing documentation, generating tests, and building features.
Jules runs asynchronously and uses Google's Gemini 2.5 Pro model, which is optimized for coding. After signing up at jules.google.com and linking a GitHub account, developers can choose a repository and branch where they want to make changes. From there, they enter a prompt like "Add a test for parseQueryString in utils.js" to get started.
According to Google, Jules breaks down complex tasks into smaller steps, adapts to user instructions, and runs unit tests to verify its work. Developers can also customize Jules for their specific project environment using setup scripts. The tool operates inside a cloud-based virtual machine. It starts by cloning the repository, installing dependencies, analyzing the codebase, and then creating a plan for testing.
Pull requests with audio summaries
Once it finishes, Jules automatically creates a pull request. Developers can review the changes, merge them, and commit them to the repo. It also adds an audio summary of the modifications—a feature that's already proven popular in Google's NotebookLM.
Jules does its work in the background, so users can close their browser while tasks are running. Progress updates and questions show up as browser notifications or within the interface.
Jules is currently available through Google Labs after registration, and access will expand regionally as capacity increases. Google is also maintaining a public GitHub repository with example prompts to help users get started.
