Argent Cloud Overview: Let Your Agents Run iOS Simulators in the Cloud

When we first started experimenting with cloud agents, we quickly hit a problem: the iOS Simulator only runs on macOS, while most cloud agents do not. Agents could write the code, but had no way to run the app and see if it actually worked. We already had Argent, our toolkit to control, debug and profile mobile applications, but for iOS the agent needs access to a Mac. We looked at the existing solutions, but none of them fit our needs.

That’s why we built Argent Cloud, a cloud simulator infrastructure for AI agents.

What Argent Cloud gives you

Argent Cloud gives you access to iOS simulators running in the cloud that you can control from anywhere, just like you would if they were running on your own machine. Your agents and CI pipelines can install your app, tap through it and take screenshots without needing to run on a Mac. The simulators run on Apple hardware that we manage for you, so they stay fast and smooth, and you don’t have to worry about maintaining the infrastructure yourself.

Each Mac is dedicated to you, never shared with other customers. It runs 4 isolated sessions in parallel, all for one flat price, so you can test your apps as much as you need.

Who it’s for

• Teams using cloud AI agents to write mobile code.
• Teams that want their CI to test on iOS without adding Macs to the pipeline.
• Cross-platform developers on non-Mac machines who want to check their app on iOS.
• Teams with complex agent workflows that run a lot of tests.

Use cases

You can’t really call an app ready to ship until you’ve seen it working on iOS. Some features are simply hard to get right without seeing them in action. A few cases where Argent Cloud comes in handy:

• Developing against local servers. Servers on your machine are tunnelled to the cloud simulator, so your app reaches them just as it would locally. With React Native, Metro is one of them, so hot reload works as usual and every change appears in the live preview right away.
• Testing camera features, such as document scanning, by feeding an image to the virtual camera.
• Letting an agent finish a feature on its own. It taps through the new flow and checks screenshots before the change reaches review.
• Catching iOS regressions on every pull request, with your existing Appium, Maestro or Argent Flow suite running on fresh, already booted simulators.

How it works

You log in from any machine, whether that’s macOS, Linux or a cloud sandbox, and get a session. It comes with a fresh, clean simulator that’s already booted, so there’s no cold start to wait through.

From there, you control it from your machine (or your agent does):

• Nearly all simctl commands are available as a drop-in for xcrun simctl, with the same output and exit codes, so your existing scripts keep working.
• You drive the simulator the way a user would, with taps, gestures and typing, along with access to screenshots and a live preview of the screen.
• A virtual camera feeds an image or video into the simulator’s camera, so you can test features like QR scanning or document capture.
• Two-way port tunnels connect the simulator and your machine, so your local development servers (e.g. Metro, if you use React Native) work with a cloud simulator and hot reload works as usual.
• Argent Cloud integrates seamlessly with Argent, so the same tools that you use locally, such as the network inspector, work the same way on cloud simulators.

Argent Cloud for AI agents

With Argent Cloud, cloud agents can verify their own work. The agent can install the app, navigate to the part that changed, check how it works and take screenshots, all before a human needs to review the change. Running on cloud agents platforms such as Cursor or Devin, it can use Argent or other tools just as it would on a local simulator.

If you prefer to keep an eye on your agents as they work, you can do so with our web UI, which shows a live preview of each simulator they’re using.

Argent Cloud for CI jobs

Running iOS tests in CI usually leaves you with two options. You can use hosted macOS runners, but they cost up to an order of magnitude more than Linux ones and often come with limited concurrency. The alternative is to run your own Macs, which you then have to maintain. With Argent Cloud you can keep your CI pipeline on Linux runners and get an iOS simulator whenever a job needs one.

It works with your current E2E testing framework, whether that’s Appium, Maestro or Argent Flow. Simulators are already booted when the job starts, so you skip the cold boot, and every run starts from a clean state, so one job can’t leave anything behind for the next.

Why flat pricing

Argent Cloud costs one flat price per Mac instead of billing per minute. Agents run long and in loops, and per-minute billing punishes exactly that kind of work. With a flat price, sessions can stay open as long as you need at no extra cost.

We’ve been building apps for years, so we know what it takes to build a reliable production app. Every feature gets checked, fixed and checked again before it ships. That’s why we chose a flat price. It suits teams that aim to build complex apps, then grow and scale them, because the bill stays the same however many runs that takes.

Argent Cloud vs GitHub macOS runners

One Argent Cloud Mac costs a flat $299 a month and gives you 4 isolated sessions. By comparison, GitHub’s standard macOS runner costs $3.72 an hour ($0.062 per minute as of October 2026). The table below compares the two for a few common workloads, assuming one simulator per GitHub runner. Workloads driven by your team’s pushes and pull requests count 21 working days a month, while nightly and around-the-clock workloads count all 30 days.

Use case	Simulator-hours a month	GitHub macOS runners	Argent Cloud
A nightly 1-hour regression suite, on iOS	60	$223	$299
An agent verifying 10 pull requests a day, 30 minutes each	105	$391	$299
A 20-minute iOS suite on every push, 25 pushes a day	175	$651	$299
One agent iterating around the clock	720	$2,678	$299
Four agents iterating around the clock	2,880	$10,714	$299

Argent Cloud becomes the cheaper option at about 80 simulator-hours a month. The free minutes in your GitHub plan can raise that point to about 160 hours on Enterprise. Below it, GitHub’s per-minute billing costs less. Above it, every extra hour is covered by the flat price, which matters most for agents that keep a simulator busy for hours. If you need more than 4 sessions at once, add another Mac for $299.

What’s next

We’re actively working on cloud builds. You’ll be able to build your iOS app straight from your project in a fresh macOS VM, and the result will install directly onto your simulator. Once that lands, Argent Cloud will take care of every step that needs a Mac, so your pipeline will go from source code to a tested app without you having to maintain a single Mac. Your agents will be able to build and test their changes entirely on their own.

Android is coming too. We’re bringing Android emulators to Argent Cloud, so the agent that checks your change on iOS will be able to check it on Android as well before the pull request reaches review. Plus, your CI will be able to run the same Appium or Maestro suite against both platforms.

Where to start

• Argent Cloud: https://argent.swmansion.com/cloud
• Argent Cloud docs: https://docs.swmansion.com/argent/docs/cloud/fundamentals/getting-started/
• Security and data isolation: https://docs.swmansion.com/argent/docs/cloud/reference/security-and-data/