Getting Started with Jules, An Asynchronous AI Coding Agent by Google

During the 2025 Google I/O, the company launched Jules, an asynchronous artificial intelligence (AI) coding agent. Before we tell you more about Jules, you need to understand what asynchronous means here. In computer programming terms, or in this particular scenario, asynchronous means the process that works independently of each other, without the program needing to pause one task to finish before starting the next.

What is Jules?

Jules is an asynchronous AI coding agent by Google that integrates directly with your existing repositories. It can clone your codebase into a secure Google Cloud virtual machine (VM), reason across the full project, understand the full context of your project, and return a pull request.

Jules is an autonomous, agentic coding system powered by Gemini 2.5 Pro that can read your code, understand your intent, and get to work. Instead of limiting itself to a prompt window, it is a highly skilled, autonomous AI agent that works in the context of real branches and Issues, running several jobs inside the same VM at once.

What Jules actually does

Jules is designed to operate asynchronously. This means you can give it a task like fixing a bug, building out a new feature, writing tests for existing code, or even updating software dependencies, and then turn your attention elsewhere.

• Full-repo context: Jules operates directly within your existing projects, understanding the complete context to make intelligent multi-file edit changes.

• Parallel execution: Because it runs in a cloud environment, it works simultaneously on tasks to fix errors, make features work, and ensure version bumps don't queue behind one another.

• Visible plan and diff: Before committing any changes, Jules shows its reasoning (what it intends to do and why) and a clean diff, letting reviewers accept, tweak, or reject modifications (the "diff") before they are finalized.

• GitHub-native workflow: It connects with GitHub, meaning developers can interact with Jules using the tools they already use daily.

• Audio changelogs: An interesting feature is the ability to generate audio summaries of recent code commits, offering a new way to catch up on project developments.

• Strict privacy: Your private code isn't used to train the general Jules model, and your data remains isolated within its execution environment (VM).

Jules has access to sophisticated reasoning abilities, allowing it to tackle complicated, multi-file changes due to being powered by Google's Gemini 2.5 Pro. When finished, it provides its plan, reasoning, and code changes for review, often preparing pull requests for easy integration.

How to use Google Jules:

Step 1: Visit Jules and click on Try Now.

Step 2: Connect your GitHub account to get started.

Step 3: Simply ask it to solve bugs, small feature requests, and other software engineering tasks.

• Jules will take some time to think and find the right answer.

Conclusion:

Jules will not replace developers, but it will change the tempo of software work. With more autonomous, agentic coding agents entering into everyday repositories, the conversation is likely to change from "Can an AI write code?" to "Which tasks still need a human touch?" For companies willing to test Jules now, the experiment offers an early look at that new balance, which is more of a collaborative effort between humans and intelligent machines.