Google now lets you code native Android apps with AI Studio

Apple may not be a fan of vibe coding, but Google is leaning into it. On Tuesday, Google announced that anyone can now build native Kotlin-based Android apps with just a few prompts using Google AI Studio. To test the results, users can install the app on their own devices and share it with others to test, or bring it into Android Studio to fine-tune the details and enable broader distribution.

As Google said in its announcement, AI has made it almost easy to build web-based applications (though running them in production is another matter). But with everyone running MVPs in the browser, mobile development has taken a bit of a backseat. localhost:3001. While Google already supports Gemini for vibe coding in mobile apps in Android Studio (and starting today, it also supports virtually all popular LLMs), there was certainly a hurdle to getting started for many non-developers.

“By combining the ease of use of a prompt-based interface with the power of the Android SDK, we now have the best of both worlds. It’s all done within the browser and requires no installation.”

“By combining the ease of use of a prompt-based interface with the power of the Android SDK, we now have the best of both worlds – all done within the browser and with no installation required,” Google writes.

The company says that building Android apps in AI Studio, Google’s web-based tool for testing new models and prototyping applications, uses the same technology as building apps with Gemini in Android Studio. The app is built in Kotlin and uses Google’s Jetpack Compose, the current standard for building Android apps.

What this actually enables is the ability to use all the features of a mobile device, including built-in sensors, GPS, Bluetooth, and NFC.

However, what this actually allows you to do is use all the features of your mobile device, including built-in sensors, GPS, Bluetooth, and NFC. Google’s example shows how this could work on the Pixel Watch (which is also an Android device, after all), using data from GPS, gyroscopes, and other sensors to build an avionics panel that mimics what’s inside a small airplane. However, Google’s prompt asks for an airspeed indicator, which the watch can’t tell (though groundspeed helps with this).

Of course, Google also notes that Gemini can be integrated into apps to provide AI capabilities.

AI Studio has a built-in Android emulator to preview your app, but the real test is how well it runs on a real device. For this, users just need to connect their Android smartphone to the machine running AI Studio, and the integrated Android Debug Bridge will take over.

To continue building apps with Android Studio, Google allows developers to download their code as a zip file.

What’s particularly interesting here is that Google is also building a direct connection to the Google Play Console. This allows users to upload apps directly to Google Play for testing and sharing. One thing to note is that you need a Google Play developer account. This is subject to a one-time $25 registration fee. “AI Studio automatically creates an app record, packages the bundle, and uploads it to an internal test track in the Google Play Developer Console.”

You’ll soon be able to share this app with friends and family directly from AI Studio.

Also on the roadmap is integration with Firebase, Google’s backend platform for adding things like databases and authentication to mobile apps.

Prior to joining The New Stack as Senior Editor for AI, Frederic was the Enterprise Editor at TechCrunch, where he covered everything from the rise of the cloud and the early days of Kubernetes to the emergence of quantum computing.