Google Jules is now generally available, and has had a 'critic' mode added to help reality check the suggestions the tool makes. 

Jules is an asynchronous, agentic coding assistant that integrates directly with existing repositories. It clones your codebase into a secure Google Cloud virtual machine (VM), understands the full context of your project, and performs tasks such as writing tests, building new features, providing audio changelogs, fixing bugs, and bumping dependency versions.

Jules can handle Python and JavaScript coding tasks. It uses Gemini  and will handle bug fixes and other time-consuming tasks. Jules creates comprehensive, multi-step plans to address issues, efficiently modifies multiple files, and even prepares pull requests.

Developers can review the plans Jules creates and provide feedback or request adjustments before deciding whether to merge the code Jules writes into their projects. Jules is private by default, it doesn't train on your private code, and your data stays isolated within the execution environment.

Jules is now available in three tiers - a basic version that you can use for 15 tasks per day and 3 concurrent tasks; a Pro tier that can be used for 100 tasks per day, which Google says is enough to run Jules throughout your coding day. The Pro tier can be used for 15 concurrent tasks, so you can run multiple threads in parallel, and offers higher access to the latest models, starting with Gemini 2.5 Pro. The basic tier also supports Gemini 2.5 Pro. 

Finally, there's an Ultra tier that supports using Jules for 300 tasks per day, and 60 concurrent tasks, built for massively parallel workflows. This tier is described as offering priority access to the latest models, starting with Gemini 2.5 Pro. 

Jules has undergone a beta trial during which the user feedback led to Google refining the interface and introducing new features. 

This release also adds a 'critic', a tool within Jules that reviews and critiques code before you see it. The aim is to catch problems of subtle bugs, missed edge cases, and untested assumptions. 

Critic augmented generation means that while Jules builds, the critic functionality looks at each proposed change in an adversarial review before completion. The critic is integrated directly into the code generation process. Google says that this first version is a one-shot process that evaluates the final output in a single pass. Future milestones will aim to make it a true multi-step agent, capable of using tool calls or triggering after subtasks or before planning, but for now it reviews the complete generation at once. This draws on multi-step research, where a critic validates outputs using external tools (like a code interpreter or search engine) and learns from those outcomes.

The critic doesn't fix code, it flags it, then hands it back to Jules to improve. Examples given by Google include a patch that introduces a subtle logic error, or a change that compiles but silently drops a required field. Another example given is that of code that works but uses an inefficient approach. 

Google says this isn't just another linter or test, because while linters follow shallow, fixed rules, the critic understands the intent and context behind code and is closer to a reference-free evaluation method. 

Google Jules is available now.

More Information

Jules Website

Related Articles

Google Releases Gemini 2 And Jules Code Agent

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info