Google’s AI Coding Assistant Jules Officially Launches Beyond Beta
On Wednesday, Google officially introduced its AI coding assistant, Jules, moving out of beta just over two months after its public preview launched in May.
Powered by Gemini 2.5 Pro, Jules is an asynchronous coding agent that integrates with GitHub, creates codebases in Google Cloud virtual machines, and uses AI to make modifications or improvements to code, allowing developers to concentrate on other tasks.
Jules was initially unveiled as a project within Google Labs in December and later became available to beta testers during its public presentation at the I/O developer conference.
Kathy Korevec, Google’s director of product at Google Labs, told TechCrunch that the tool’s improved stability was the reason for its transition out of beta, following extensive UI and quality enhancements during that phase.
“The direction we are taking gives us significant confidence that Jules is here to stay,” she noted.
With the wider release, Google has rolled out structured pricing for Jules, starting with a free “introductory access” tier limited to 15 individual daily tasks and three concurrent tasks, down from the 60-task limit of the beta period. Paid plans are available through Google AI Pro and Ultra, priced at $19.99 and $124.99 per month, offering users 5× and 20× greater limits respectively.
Korevec emphasized that Jules’ pricing model is based on “real usage” data collected over the preceding months.
“The 60-task limit allowed us to evaluate how developers interacted with Jules and provided insights necessary to shape the new pricing model,” she shared. “The 15-task limit is designed to help users determine whether Jules meets their needs on actual projects.”
Furthermore, Google has updated Jules’ privacy policy for clearer communication regarding its AI training practices. If a repository is public, its data may be used for training, whereas for private repositories, Korevec confirmed that no data is sent out.
“We received feedback indicating that the privacy policy wasn’t as transparent as we had hoped, so most of the updates respond to that. We did not change our training practices but clarified the language,” Korevec explained.
During the beta, Google noted that thousands of developers completed tens of thousands of tasks, resulting in over 140,000 publicly shared code improvements. Initial feedback led the Google Labs team to add new features like reusing previous setups for faster task execution, integration with GitHub issues, and support for multimodal inputs.
The primary users of Jules so far have been AI enthusiasts and professional developers, according to Korevec.
Operating asynchronously within a virtual machine, Jules sets itself apart from leading AI coding tools like Cursor, Windsurf, and Lovable, which function synchronously and require users to monitor outputs after each action.
“Jules acts as an extra set of hands… you can initiate tasks and then step away from your computer, returning later to find them finished, unlike a local agent or synchronous tool that binds you to the session,” Korevec detailed.
This week, Jules received improved integration with GitHub to automatically create pull requests — similar to its branch functionality — and a feature called Environment Snapshots, which allows it to save dependencies and installation scripts as snapshots for faster and more consistent task execution.
From vibe coding to mobile use, beta trials informed Jules’ development
Since its public beta launch, Jules has recorded 2.28 million visits globally, with 45% coming from mobile devices, based on data from market intelligence provider SimilarWeb reviewed by TechCrunch. India ranks first for traffic, followed by the U.S. and Vietnam.
Google has not shared specific information about Jules’ user demographics or leading regions.
Korevec mentioned to TechCrunch that during the beta, the team observed many users utilizing Jules to fix bugs or enhance vibe-coded projects to better prepare them for production.
Initially, Jules required users to have an existing codebase. However, Google soon realized that many potential users, particularly those experimenting with various AI tools, wanted to test the tool without one. Korevec noted that enabling access to Jules even with an empty repository significantly expanded its applicability and usage.
The Google Labs team also noted a rising number of users accessing Jules through mobile devices. While there isn’t a dedicated mobile app, Korevec stated that users are leveraging its web app for access.
“Given that we’re seeing this emerge as a significant use case, we are actively looking into what features mobile users need,” she pointed out.
Alongside beta testers, Korevec indicated that Google is already utilizing Jules for developing various internal projects, and a significant initiative is underway to integrate the tool in “many more projects” within the company.
