At I/O 2025, Google unveiled Jules, a global release of its autonomous coding agent. Jules can analyze your codebase, grasp your intent, and take action accordingly. It operates asynchronously by connecting to your existing repository and securely cloning the codebase into a Google Cloud virtual machine (VM).
This means the AI agent operates entirely in the cloud rather than on your local machine. Much like OpenAI Codex, Google’s Jules can write tests, develop new features, fix bugs, and update dependencies. Additionally, it offers audio changelogs for easier review.
You can assign tasks to the Jules coding agent, which works autonomously in the background. Once a task is complete, it presents its reasoning, plan, and a diff of the changes made. Google assures that your private codebase is never used to train AI models, and all data stays isolated within a secure environment.
Google has also introduced parallel execution for Jules, allowing you to assign multiple tasks at the same time. You can connect your GitHub repo so Jules fully understands your project context. Currently in public beta, Jules is free to use with some usage limits. Access it anytime at jules.google.com.
