Question 1

What is Qalti?

Accepted Answer

Qalti is a QA-automation app that drives the standard iOS Simulator with a large-language-model (LLM) “virtual user.” It lets you record or write test scenarios in plain English and then re-run them automatically. No source-level hooks or test SDKs required: everything is done through computer-vision-powered interactions.

Question 2

Who is Qalti for?

Accepted Answer

Anyone who tests iOS apps. Solo developers that want to spend more time building and less time context-switching. Mobile dev teams who need functional regression coverage on iOS but don’t want to maintain brittle coordinate-based scripts or XCUITest code. Early adopters have been QA engineers and indie developers, but PMs/designers can write and tweak scenarios too.

Question 3

What types of issues can you detect?

Accepted Answer

We detect UI issues, performance problems, and functional bugs across both iOS and Android platforms. Our AI tests on real devices to catch issues that might be missed in emulators.

Question 4

What do I need to run it?

Accepted Answer

• macOS 13 or later (Apple silicon recommended for speed)
• Xcode with the iOS Simulator installed
• An invitation from the private-alpha wait-list

Question 5

Does it run on Windows or Linux?

Accepted Answer

Not yet. macOS is required because we piggy-back on Apple’s Simulator APIs.

Question 6

Can I test on a real iPhone or iPad?

Accepted Answer

Not today. Physical-device support (USB-tethered first, cloud device farm later) is on the near-term roadmap.

Question 7

How do I create a test?

Accepted Answer

1. Launch the Simulator and use the app’s “Record” mode—every tap, swipe, and text entry is logged.
2. Stop recording to generate a natural-language script.
3. Edit the script in plain English or dive into the advanced markup for precise actions (e.g., “tap ‘Login’ button,” “type ‘hello@example.com’ into field labeled ‘Email’”).

Question 8

Does it handle custom UIs (Unity, Unreal, Flutter, etc.)?

Accepted Answer

Yes—because decisions are purely visual, any pixels on screen are targetable, including game engines and OpenGL/Metal surfaces.

Question 9

Do screenshots leave my machine?

Accepted Answer

During the alpha, model inference happens in the cloud; screenshots are transmitted over TLS, stored transiently, and deleted after each run. On-device model execution is planned for a later milestone.

Articles