🚀 Try It Now → jules.google.com
Jules is Google's autonomous coding agent — live now, free with your Google account. Here's how to get in:
- Go to jules.google.com and sign in with Google
- Connect your GitHub account — choose which repos Jules can access
- Pick a repo + branch and write a goal (not a prompt — a target outcome)
- Click "Give me a plan" — Jules generates a plan for your approval before touching any code
- Review, approve, and let it run — you'll get notified when it's done or needs input
💡 Pro tip: Add an AGENTS.md file to your repo root. Jules reads it to better understand your codebase and generate smarter plans.
⚠️ Note: Jules is currently in public beta. Access is free during beta with usage limits — pricing has not been announced yet. Availability may vary by region.
Get Started — How to Think in Goals, Not Prompts
Jules is goal-based, not prompt-based. Here's how to hit the ground running:
- Think in outcomes, not instructions. Instead of "write a function that does X", try "the auth flow needs to handle token expiration gracefully."
- Start small. Give Jules a well-defined goal in a contained part of your codebase before handing it something critical.
- Use the checkpoints. When Jules flags a decision for your approval, engage at the architectural level — don't rubber-stamp it.
What Makes Jules Different
Every AI coding tool until now runs the same loop: you prompt, it executes, you adjust, repeat. Jules flips that. You define a goal and it figures out the path, the steps, and the execution — running inside a persistent workspace with full context across your codebase.
That's not smarter autocomplete. That's autonomous reasoning.
Key Features
- 🎯 Goal-based input — describe outcomes, not instructions
- 🔗 GitHub Integration — imports repos, branches changes, helps create PRs
- ☁️ Virtual Machine — clones your code in a Cloud VM and verifies changes work
- 🧪 Test Suite — runs existing tests or creates new ones automatically
- 🌐 Available Anywhere — CLI access or build your own workflows via API
- ✅ Approval Checkpoints — human-in-the-loop at every meaningful decision point
Why Oversight Matters
When AI starts deciding what code should exist, stakes change. Google built approval checkpoints into Jules's core — giving you visibility before execution and control to redirect or halt at any point. Autonomous capability without oversight is a liability. Jules gets that right.
🏄 The AI Surfer — built for people who want to work with agents, not just react to them. Dallas built this community around one idea: automate the 90% you hate, own the 10% you love. theaisurfer.com →