Testing

Nobody likes writing tests. But everybody likes having them. AI coding tools have gotten surprisingly good at generating tests — and surprisingly bad at certain parts of it. The teams winning with AI in 2026 are not generating the most code. They are building processes to ship reliable code despite the fact that AI introduces roughly 1.7 times more bugs than humans, according to recent studies. Good test generation is a critical part of that process. ...

You have shared code running on Android and iOS. But how do you know it works correctly on both platforms? You test it. KMP has built-in support for testing. Tests written in commonTest run on every platform automatically. Write once, verify everywhere. Test Source Sets KMP projects have test source sets that mirror the main source sets: shared/src/ ├── commonMain/ → shared code ├── commonTest/ → tests for shared code (runs on all platforms) ├── androidMain/ → Android-specific code ├── androidTest/ → Android-specific tests ├── iosMain/ → iOS-specific code └── iosTest/ → iOS-specific tests Tests in commonTest are the most valuable — they verify your shared logic on every target platform with a single test file. ...

Your app works when you manually tap through it. But does it work after your next code change? And the one after that? Manual testing doesn’t scale. You can’t tap through 50 screens after every change. That is what automated tests are for — and Compose makes testing surprisingly easy. Why Test Compose UI? Three reasons: Catch bugs before users do — tests run in seconds, not minutes of manual tapping Refactor with confidence — change code, run tests, know nothing broke Document behavior — tests show WHAT the UI should do Setup Add the testing dependencies to app/build.gradle.kts: ...

Writing tests is the thing most developers know they should do but often skip. It takes time. It is boring. And when deadlines are tight, tests are the first thing to go. AI changed that. In 2026, AI tools can generate comprehensive tests for your code in seconds — often catching edge cases you would miss yourself. But how good are these tests really? Can you trust them? And which tools should you use? ...