Jules is Google's asynchronous coding agent that clones a repository, plans changes, writes code and opens a pull request without a person watching it work. PR and comms teams use Jules the same way engineering teams do: to build and maintain small internal tools while the team focuses on other work.
Last updated November 5, 2026. Pricing, task limits and usage figures come from the sources linked in each section and should be checked against Google's current plans before a purchase decision.
What is Jules?
Jules is Google's autonomous coding agent, introduced as a Google Labs project in December 2024, that takes a task, clones the target repository into an isolated cloud virtual machine, writes a plan, and executes multi-file code changes before opening a pull request for review, according to Google's own product announcement.
Jules runs on Gemini models and offers tiered daily task limits: a free tier, a Pro plan and an Ultra plan with 15, 100 and 300 tasks per day respectively, per the official Jules product site. During its beta period, developers using Jules generated more than 140,000 publicly shared code improvements, according to an independent review of the platform.
How is Jules different from Antigravity?
Jules is different from Antigravity, covered separately in the Antigravity in public relations guide, because Jules works asynchronously in the background on a task and hands back a finished pull request, while Antigravity's Manager surface keeps a person orchestrating and observing multiple agents in something closer to real time.
Why it works: Jules produces a step-by-step plan before making any changes, and a person can review, edit or reject that plan before Jules executes it. That is the specific mechanism that lets a non-developer assign a task with confidence: the plan itself is readable and gives the reviewer a chance to redirect the agent before code gets written, according to Google's own description of the workflow.
Two other Google tools in the same series cover different jobs: the Gemini Agent guide for task automation and the Veo guide for video.
How would a comms team use a coding agent built for developers?
A comms team would use Jules for the same class of small internal tool covered in the Antigravity guide, a media-monitoring tracker or an approval-routing page, but through a GitHub-issue workflow rather than a conversational Manager surface, which fits a team that already has a lightweight GitHub repository for its internal tools.
Illustrative scenario: a comms team maintaining a simple internal newsroom archive site notices a broken link-checking script. Instead of pulling a developer off another project, a team member opens a GitHub issue describing the fix, assigns it to Jules, and reviews the resulting pull request once Jules finishes, the same review step any code change from a human contributor would get.
How do you write a task that Jules can finish?
A task Jules can finish names one tool, one problem and one way to check the result. Jules works from a task description and an existing codebase, so a vague request gives the reviewer a vague plan to evaluate.
- Name the tool and the file or page where the problem appears.
- Describe the current behavior and the behavior the team wants, in plain sentences.
- State how the team will know the fix worked, such as a link report that shows no broken links.
- Keep each issue to one change, so the pull request stays small enough to review in one sitting.
- Read the plan Jules writes before it executes, and edit or reject the plan if it touches files the issue did not mention.
Teams that already publish through a content workflow can apply the same discipline described in the AI PR strategy guide: define the outcome first, then measure it afterward.
What does a comms team need in place before assigning Jules a task?
A comms team needs an existing GitHub repository for the tool in question and someone able to review a pull request's diff before merging it, since Jules produces code for review, not code that deploys itself automatically.
Jules also supports an Environment Snapshot, a saved record of dependencies, install scripts and environment variables, which Jules reuses for faster and more consistent setup on repeat tasks, according to an independent technical comparison of the platform. Setting one up the first time a comms team hands Jules a task on a given tool saves setup time on every task after that.
What are the real limits of Jules for a non-engineering team?
Jules works from a task description and an existing codebase, so a comms team with no repository at all for its internal tools has nothing for Jules to act on. A tool like Opal or Antigravity's Manager surface is a better starting point for a team building its very first internal tool from scratch.
Jules also does not remove the review step. A person still has to read each pull request, because Jules produces code for review and not code that deploys itself. Teams with no one who can read a diff should settle that before the first task.
What does Jules access cost?
Jules is free for the base tier, with 15 tasks per day, and paid tiers add higher volume and faster model access: the Pro plan runs 19.99 USD per month for 100 tasks daily, and the Ultra plan runs 124.99 USD per month for 300 tasks daily, according to the official Jules product site.
For a comms team, the free tier is usually enough to test whether Jules fits a specific maintenance workflow before a department pays for a higher tier, since 15 tasks a day covers most teams' actual volume of small internal-tool fixes.
| Situation | What Jules does | What still needs a human |
|---|---|---|
| Existing internal tool needs a fix or a feature | Plans the change, writes the code, opens a pull request | Reviewing and merging the pull request |
| Repeat maintenance tasks on the same codebase | Reuses a saved Environment Snapshot for faster setup | Confirming the snapshot still matches the current environment |
| No existing codebase yet | Nothing to act on without a repository | Starting the project in a tool like Opal or Antigravity first |
What should comms teams take from Jules?
Jules extends the idea covered in the Antigravity guide, an agent that writes and tests code on a comms team's behalf, into a background, task-based workflow built around GitHub pull requests rather than a live Manager surface. That makes it the right fit once a team already has a small internal tool running and just needs ongoing fixes and features, with a human still reviewing every pull request before it merges.
Teams weighing tools like this alongside their wider technology communications plans can see how 5W approaches that work on the technology PR practice page.
Frequently asked questions
What is Jules in public relations?
Jules is Google's asynchronous coding agent. Comms teams use it to build and maintain small internal tools, such as a media-monitoring tracker, by assigning tasks through GitHub issues.
How is Jules different from Antigravity?
Jules works in the background on a task and returns a pull request, while Antigravity's Manager surface keeps a person orchestrating and observing multiple agents in closer to real time.
Does a comms team need a developer to use Jules?
A team needs a GitHub repository for the tool and someone able to review a pull request before merging it. Jules does not deploy code by itself.
How much does Jules cost?
According to the official Jules product site, the base tier is free with 15 tasks per day, the Pro plan runs 19.99 USD per month for 100 tasks daily, and the Ultra plan runs 124.99 USD per month for 300 tasks daily.
What is an Environment Snapshot in Jules?
An Environment Snapshot is a saved record of dependencies, install scripts and environment variables that Jules reuses for faster and more consistent setup on repeat tasks.
