Google launched its AI coding agent, Jules, out of beta on Wednesday, just over two months after its public preview debut in May. Powered by Gemini 2.5 Pro, Jules is an asynchronous, agent-based coding tool that integrates with GitHub, clones codebases into Google Cloud virtual machines, and uses AI to fix or update code while developers focus on other tasks.
Initially announced as a Google Labs project in December, Jules was made available to beta testers during the I/O developer conference. Kathy Korevec, director of product at Google Labs, explained that improved stability led to the decision to exit beta after hundreds of UI and quality updates. She expressed confidence in Jules’ long-term viability, stating, “The trajectory of where we’re going gives us a lot of confidence that Jules is around and going to be around for the long haul.”
With the wider rollout, Google introduced structured pricing tiers. The free “introductory access” plan now caps at 15 individual daily tasks and three concurrent ones, down from the 60-task limit during beta. Paid tiers are part of the Google AI Pro and Ultra plans, priced at $19.99 and $124.99 per month, offering 5x and 20x higher limits, respectively. Korevec noted that the new packaging and pricing were based on real usage insights gathered during beta testing.
Google also updated Jules’ privacy policy to clarify AI training practices. Public repository data may be used for training, but private repositories remain untouched. Korevec explained that the changes were made in response to user feedback, emphasizing that no actual training practices were altered—only the language was refined.
During the beta phase, thousands of developers completed tens of thousands of tasks, resulting in over 140,000 publicly shared code improvements. Feedback led to new features, including faster task execution through reused setups, GitHub issue integration, and multimodal input support.
Jules differentiates itself from other AI coding tools like Cursor, Windsurf, and Lovable by operating asynchronously in a virtual machine. Korevec described it as “an extra set of hands,” allowing users to assign tasks and return later to completed work, unlike synchronous tools that require real-time monitoring.
Recent updates include deeper GitHub integration for automatic pull requests and a feature called Environment Snapshots, which saves dependencies and install scripts for faster, more consistent execution.
Since entering public beta, Jules has recorded 2.28 million visits worldwide, with 45% coming from mobile devices. India, the U.S., and Vietnam were the top markets. Korevec noted that many beta users employed Jules to refine “vibe-coded” projects, transitioning them to production-ready states. Initially requiring an existing codebase, Google later enabled Jules to work with empty repositories, broadening its appeal.
Mobile usage has also grown, with users accessing Jules via the web app. While no dedicated mobile app exists yet, Korevec confirmed that mobile-friendly features are under exploration.
Internally, Google has begun using Jules for some projects, with plans to expand its adoption across more teams. The tool’s development continues to be shaped by real-world testing and user feedback.