Google Jules is an asynchronous AI coding agent that connects to your GitHub repository, works on tasks in a cloud virtual machine, and opens pull requests you review before merging. It launched in public beta in 2025 and moved to paid tiers in 2026, with a free daily quota for lighter use.
What is Google Jules?
Jules is not an autocomplete. It is a delegated agent. You describe a task, Jules clones your repo into a sandboxed VM, edits files, runs tests, and hands back a diff as a pull request. That async model means you assign work and walk away rather than babysitting a chat window. It is powered by Gemini models, and Google positions it against GitHub Copilot’s coding agent and OpenAI’s Codex cloud agent.
Source access is GitHub-only at launch. If your code lives in GitLab or Bitbucket, you are out of luck for now.
Product Features
- Cloud VM execution: Jules runs your build and test commands in an isolated environment, so a broken run never touches your machine.
- Pull request workflow: every task shows up as a real PR with a diff, not a paste-in-chat blob.
- Repo-wide context: it reads multiple files, not just the one you have open.
- Plan-then-execute: Jules proposes an approach first and waits for you to approve it on larger tasks.
- Gemini backend: tuned specifically for code, with Google’s latest coding models.
Product Highlights
The async framing is the real differentiator. Tools like Cursor and Cline expect you at the keyboard. Jules is built for “fix this bug while I’m in a meeting.”
The PR-first output also fits teams. You get review, blame, and CI for free because the change lands through the normal GitHub pipeline. That beats a chat window you have to copy-paste from.
Use Cases
- Clearing a backlog of small bug fixes and dependency bumps.
- Adding tests to a module that has none.
- Refactoring a function across several files while you work on something else.
- Updating docs and config files that no one wants to touch by hand.
- Prototyping a feature branch overnight and reviewing the diff in the morning.
Is Google Jules free?
There is a free tier with a limited number of tasks per day (Google has changed the exact number during beta, so check the console). Paid plans add higher daily task limits and priority compute. Google has also bundled Jules access into some Google AI Pro and Ultra subscriptions.
Jules is a poor fit if you need instant inline suggestions or work outside GitHub. It shines when you have clearly scoped tasks you can describe in a sentence.
Quick Summary
- Google Jules is an async AI coding agent that works in a cloud VM and opens pull requests.
- It supports GitHub only; no GitLab or Bitbucket at launch.
- It has a free daily task tier plus paid plans with higher limits.
- Best for delegated, scoped tasks; bad for real-time inline editing.
- It is powered by Gemini coding models from Google.
