Hi Product Hunt, I'm Lukas, I build Tapground. I was building a financial app with a team of 5 (including 1 QA) and, and every release was tested by hand because the Appium suite cost more to maintain than it caught. The test knows where the login button is. Then someone moves it, or Android lays the screen out differently, and the test goes red while the app is fine. So you fix the selector. Next week you fix another one. A year in you have two suites, one per platform, and somebody whose job is mostly keeping them green. I got tired of being that somebody. With Tapground you write the test the way you'd explain it to a colleague, as YAML in plain English. "Tap the login button." "Check that the order total is 42." On every run a vision model looks at the screen and finds the thing you meant, so there are no selectors in the test and nothing to add to your app. One scenario runs on iOS Simulators and on Android Emulators. You can watch it live, and afterwards there's a video and a screenshot for every step. CI is one command. It also grew a recorder along the way. You tap through your app in the live preview and the taps come back as a scenario you can edit. A test can grab a value off the screen and use it later, and it can wait for a verification email inside the run, so a sign-up flow goes all the way through instead of stopping at the inbox. One thing I learned the hard way. During a demo the model typed an email address into the search box, because it had named the wrong field and the executor believed it. Now the executor checks what the model claims against the screen before it acts. Don't trust the model, check it. The first 50 minutes are free. Run it on your own login flow and tell me where it breaks. A failed step tells me more than a thumbs up.

Tapground

Mobile UI tests in plain English, no selectors to maintain

Mobile UI tests in plain English, no selectors to maintain