Google Jules is shutting down. Google posted the notice in the official Jules changelog on October 29, 2026. Jules will wind down on December 31, 2026, and it turns read-only after November 11, 2026. If your team uses this AI coding agent, you have only a few days left for active work.

I'm Nitin, and I coordinate projects for our dev teams at Debugged Pro. When a tool in the delivery chain goes away, the risk is rarely the code. The risk is the setup, the habits and the small jobs nobody wrote down. This guide covers the verified facts and a simple plan to move on.

Google Jules asynchronous coding agent thumbnail image from Google — Image credit: Google, from the Google Keyword post introducing Jules as an asynchronous coding agent.

What Google Announced About Jules

The notice is short and clear. Google says the move is part of its work to unify its agentic coding efforts. Here is the timeline from the official Jules turndown notice:

Date	What happens	What you can do
October 29, 2026	Shutdown announced	Everything, including export
Until November 11, 2026	Normal service	Finish sessions and export data
November 11 to December 31, 2026	Read-only	View past work
December 31, 2026	Jules winds down	Nothing inside Jules

Google calls Jules an experiment. It says the project taught the team how asynchronous coding agents can help people clear a backlog and ship faster. Those lessons are now shaping other Google products, including Google Antigravity, its agent-first platform.

Google also shared a note on X, quoted by The Verge. It said the team is bringing its learnings from Jules into Antigravity to offer "autonomous remote background task execution and an agent-first development environment."

A Quick Look Back at Google Jules

Jules first showed up in Google Labs in December 2024. It entered public beta in May 2025. Google then took it out of beta in August 2025, powered by Gemini 2.5 Pro.

The idea was simple. You pointed Jules at a GitHub repo and gave it a task. It cloned the code into a cloud virtual machine, made a plan, did the work and handed back a diff or a pull request. You did not have to sit and watch it.

Google's launch post on the Keyword blog said beta users produced over 140,000 code improvements shared publicly. Paid plans came through Google AI Pro and Ultra, with 5x and 20x higher limits.

Google Jules screen showing a coding task plan waiting for user approval — Image credit: Google. Jules shows its plan before it changes code, and the user can approve or edit it.

That plan-first flow is worth remembering. It is a good habit for any AI coding agent, and you should look for it in whatever tool you pick next.

Why the Google Jules Shutdown Matters for Agencies

Most agencies did not run their whole pipeline on Jules. Still, many teams in India and the USA tried it for side work. Think dependency bumps, test fixes and small backlog tickets.

Those small jobs are exactly what slips through during a shutdown. Here is why it matters:

Scheduled tasks can stop quietly. If Jules ran a weekly job for you, nobody may notice when it ends.
Setup knowledge lives inside Jules. Setup scripts, environment variables and repo memories are stored there, not in GitHub.
Client trust is on the line. If a client expects automated updates, a silent stop looks like a missed promise.
Secrets need care. The export includes your environment variable values, so the files must be handled like passwords.

The bigger lesson is about tool risk. AI coding tools are changing fast. A tool you rely on today can be merged, renamed or closed within two years.

What Stays in GitHub and What You Can Lose

Google is clear on this point. Your code, branches and pull requests stay in GitHub. Nothing Jules already pushed will vanish.

What lives only inside Jules is the configuration and history around that code. Per the notice, the export covers:

Account settings
Environment variables
Scheduled tasks
Connected repositories
Memories
Setup scripts
Every session, with its prompt, plan, full activity log and code patches

For an agency, the session history can be useful. It shows what the agent changed and why. That can help with client audits or later bug hunts.

How to Export Your Google Jules Data

Google published an export path in the turndown notice. It uses Jules Tools, the command line client. Here are the steps in plain terms:

Install Jules Tools. Run npm install -g @google/jules.
Sign in. Run jules login. It opens a browser sign-in.
Connect GitHub first. If you used Jules with GitHub repos, make sure the account is linked. Otherwise the export will skip repo data.
Run the export prompt. Open a CLI coding agent, such as Claude Code, from your home directory. Paste the export prompt from the notice.
Check the folder. Your data lands in a local jules-export folder in your home directory.

A few practical notes from the notice itself:

The access token expires about an hour after login. If calls start failing with 401 errors, sign in again.
Some of the endpoints the script uses are undocumented. That is one more reason to export early.
The file env-variables.json holds real values. Keep the export folder private and never commit it to a repo.

If your agency has several team accounts, assign one owner per account. Ask each person to confirm the export with a quick screenshot of the counts the script prints.

A 7-Step Migration Plan for Agencies

Here is the plan we would follow for a small or mid-size dev team. It fits in a single sprint.

List every Jules user. Ask the team who logged in during the past six months. Include interns and contractors.
Find recurring jobs. Check scheduled tasks first. Write down what each one did and how often.
Export before November 11. Do not wait for the read-only phase. Export while the service is fully active.
Move secrets to a vault. Put environment values into your secret manager. Then delete the local copy you do not need.
Save setup scripts in the repo. Add them to a docs or scripts folder so the next tool, or the next hire, can use them.
Pick and test a new agent. Run the same three real tasks on two candidate tools. Compare quality, review effort and cost.
Tell affected clients. If any client work depended on Jules, send a short note on what changes and when.

Step five is the one most teams skip. Yet it is what makes you less dependent on any single vendor.

Choosing What Comes After Google Jules

Google points users toward Antigravity. But Google has not said Antigravity is a one-to-one swap for every Jules feature. So treat it as one option, not the default.

The market has plenty of choices. Coverage of the shutdown names rivals like OpenAI Codex, Claude Code and GitHub Copilot. Whatever you choose, judge it on a few basics:

Plan before code. Does it show a plan you can approve or edit?
Background work. Can it run tasks while your team does other things?
Repo access controls. Can you limit it to specific repos and branches?
Data handling. Does the vendor state clearly that it does not train on your private code?
Exit path. Can you export your settings and history if this tool also shuts down?

Also think about review load. AI agents can write code faster than people can check it. We covered this in our post on the AI coding agents review bottleneck, and it applies to any Jules replacement.

Security deserves a look too. Fake tools and extensions often pop up when users are searching for alternatives. Our recent guide on the fake Claude Code extension supply chain attack shows why you should install agents only from official sources.

Lessons From the Jules Shutdown for Your AI Tool Stack

Jules lasted less than two years from Labs preview to wind down. That is normal for AI products right now. Plan for it instead of being surprised by it.

A few habits help a lot:

Keep config in your own repos. Setup scripts, agent instructions and test commands belong in version control.
Keep a tool register. Track every AI tool, its owner, its plan and what data it holds.
Avoid single points of failure. Make sure any task an agent runs can also be done by a person or a second tool.
Review tools every quarter. Check for pricing changes, policy changes and product news.

Want help setting up a safer AI-assisted workflow? Our software development consulting team helps agencies and businesses plan this kind of change.

Key Takeaways

Google Jules becomes read-only after November 11, 2026 and shuts down on December 31, 2026.
Your code, branches and pull requests stay in GitHub.
Settings, environment variables, scheduled tasks, memories, setup scripts and session history must be exported.
Export with Jules Tools before November 11 and keep the files private.
Google is folding lessons from Jules into Antigravity, but test any replacement on real tasks first.

Final Thoughts

The Jules shutdown is not a crisis for most teams. It is a deadline. Spend an hour this week to find every Jules user, export the data and move recurring jobs.

Then use the moment to make your AI tool stack easier to change. The next shutdown will come, and you will be ready for it.

About the Author

Nitin Kumar is the Lead Project Coordinator at Debugged Pro. He helps dev teams in India and the USA plan delivery, choose tools and keep client projects on track. He writes about AI tools, software teams and practical ways to ship better work.