Google's Jules: The Autonomous Coding Agent That Works While You Don't

Google has officially released Jules in public beta, making their autonomous coding agent available worldwide. Unlike code completion tools or co-pilots, Jules is a fully autonomous agent that reads your code, understands your intent, and works independently on tasks you'd rather not handle yourself.

Jules operates asynchronously in a secure cloud environment, allowing you to focus on other work while it handles tasks like writing tests, fixing bugs, building features, or updating dependencies. When finished, it presents its reasoning and a diff of the changes made.

Key Features and Capabilities

Jules leverages Gemini 2.5 Pro, giving it access to advanced coding reasoning capabilities. Combined with its cloud VM system, it can handle complex, multi-file changes with impressive speed and precision.

Jules represents Google's latest effort to corner the market for AI agents, which are widely regarded as a more practical and profitable form of chatbot. The agent is built on Gemini 2.5 Pro, which has outscored other industry-leading models on key benchmarks like math and code editing.

Getting Started with Jules

Getting started with Jules is straightforward. You'll need to connect your GitHub repository and configure your environment.

Once connected, you can start assigning tasks to Jules using natural language prompts. For example, you might ask it to:

• "Fix the failing tests in the authentication module"
• "Update all dependencies to their latest versions"
• "Implement a dark mode toggle for the UI"
• "Refactor the user service to use async/await instead of promises"

Package Manager Configuration

Jules doesn't properly detect alternative package managers like pnpm and Bun automatically. You need to configure these manually using custom setup scripts in the Configuration tab of your Jules project:

set -ux 
export CI=true
cd /app

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
export NVM_DIR="$HOME/.nvm"
if [ -s "$NVM_DIR/nvm.sh" ]; then
  set +e; \\. "$NVM_DIR/nvm.sh"; NVM_SOURCE_STATUS=$?; set -e
  if [ $NVM_SOURCE_STATUS -ne 0 ] && [ $NVM_SOURCE_STATUS -ne 3 ]; then
    echo "ERROR: NVM sourcing failed: $NVM_SOURCE_STATUS"; exit 1
  fi
else
  echo "ERROR: NVM script not found"; exit 1
fi
set -e 

nvm install 22 && nvm use 22 && nvm alias default 22
npm install -g pnpm@10
pnpm install --frozen-lockfile

set -ux
export CI=true
cd /app

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
export NVM_DIR="$HOME/.nvm"
if [ -s "$NVM_DIR/nvm.sh" ]; then
  set +e; \\. "$NVM_DIR/nvm.sh"; NVM_SOURCE_STATUS=$?; set -e
  if [ $NVM_SOURCE_STATUS -ne 0 ] && [ $NVM_SOURCE_STATUS -ne 3 ]; then
    echo "ERROR: NVM sourcing failed: $NVM_SOURCE_STATUS"; exit 1
  fi
else
  echo "ERROR: NVM script not found"; exit 1
fi
set -e 

nvm install 22 && nvm use 22 && nvm alias default 22
curl -fsSL https://bun.sh/install | bash
export BUN_INSTALL="$HOME/.bun"
export PATH="$BUN_INSTALL/bin:$PATH"
bun install

These scripts:

• Install NVM (Node Version Manager)
• Set up Node.js 22 (latest stable)
• Install the latest version of your chosen package manager
• Install project dependencies with frozen lockfiles

Both scripts are minimal but include necessary error handling for the Jules cloud environment.

The AI Coding Agent Landscape in 2025

The release of Jules comes amid intense competition in the AI coding agent space. According to VentureBeat, we're witnessing a battle for the AI developer stack, with several major players offering competing solutions.

Here's how Jules compares to other major AI coding assistants:

• OpenAI Codex (Introducing Codex): Recently released as a research preview, Codex offers a polished UI with inline diff views and a mobile app. However, it currently lacks direct network access within requests and two-way GitHub sync that Jules provides.

• GitHub Copilot Agent: Building on the popular Copilot, this agent now offers asynchronous code testing and generation. While it integrates well with GitHub (naturally), it doesn't have the same level of VM-based execution environment as Jules.

• Devin (devin.ai): One of the first autonomous coding agents, Devin offers two-way GitHub integration and real-time preview deployments on Vercel. Its IDE-like controls allow engineers to iterate naturally via Slack or its native UI.

• Manus (manus.im): Specializes in code reviews and PR workflows, excelling at rapid context switching and fetching relevant code references within active pull requests.

When choosing between these tools, consider your specific workflow needs—such as network access requirements, GitHub integration depth, UI preferences, and CI/CD integration capabilities.

Trying Jules Today

During the public beta phase, Jules is available for free with some usage limits. According to Google, they expect to introduce pricing after the beta as the platform matures.

For comprehensive documentation on Jules, including detailed guides on running tasks, reviewing code diffs, and managing environments, visit the official Jules documentation.

Have you tried Jules or other AI coding agents? I'd be interested to hear about your experiences, especially if you've used alternative package managers with these tools.