Jules is an autonomous AI coding agent from Google. Unlike previous AI tools that functioned as “copilots” within an IDE, Jules operates asynchronously: the user defines a task, and the agent executes it independently within a cloud-based virtual machine. Launched in beta in late 2025 and widely available in 2026, it can analyze entire codebases, plan multi-file changes, run tests, fix bugs, and ultimately submit a complete Pull Request.
Key Features
- Multi-Agent Architecture: Comprises specialized agents for planning, execution, and critique to ensure code quality before submission.
- Asynchronous Execution: Users can assign a task to Jules and walk away; the system sends a notification once the task is complete or if input is required.
- Cloud-Native Environment (VM): Each task runs in a secure, isolated sandbox pre-loaded with development tools like Node.js, Python, Go, and more.
- Audio Changelogs: A unique feature that generates a spoken summary of the changes made, helping developers quickly understand the underlying logic.
- GitHub Integration: Connects directly to repositories, utilizes AGENTS.md files for project conventions, and creates detailed PRs with explanations.
Main Use Cases
- Software Engineers: Delegating “maintenance” tasks like dependency updates, unit testing, and performance optimization.
- Agile Development Teams: Handling small bugs in the backlog without interrupting the flow of new feature development.
- Startups and Entrepreneurs: Building entire features from scratch based on simple natural language specifications.
- Open Source Maintainers: Automatically reviewing and fixing community-submitted code contributions.
- Security Professionals: Running daily security scans and automatically fixing critical vulnerabilities.
Pricing Plans
- Jules Standard (Free): Up to 15 tasks per day (rolling 24 hours) and 3 concurrent tasks. Powered by Gemini 2.5 Pro.
- Jules in Pro: Included in the Google AI Pro subscription. Up to 100 tasks per day and 15 concurrent tasks. Access to Gemini 3 Pro.
- Jules in Ultra: Included in the Google AI Ultra subscription. Up to 300 tasks per day and 60 concurrent tasks.Priority access to the latest models.
- Enterprise Version: Custom pricing featuring unlimited tasks, enhanced security, and centralized management for dev teams.
Pros and Cons
Pros:
- Full autonomy saves hours of work on routine coding tasks.
- Ability to understand entire project context, not just single files.
- Enhanced security through execution in isolated cloud sandboxes.
Cons:
- Requires clear and specific prompts for accurate results.
- Task-based daily limits can be restrictive on lower-tier plans.
- Audio summaries and advanced features are currently optimized for English.
FAQ
- Does Jules learn from my private code? No, Google states that code from private repositories is not used to train their general models.
- Can it run my app? Yes, it can spin up servers in the VM and perform UI testing autonomously.
Screenshots and Demo
The web interface at Jules features a clean layout with a repo selector. The central command box is where users enter prompts. Once a task begins, an “Agent Dashboard” appears, showing a workflow graph: Planning, Execution, and Testing. Users can watch the terminal run in real-time and listen to the audio summary upon completion.