Build Your First Android App with Gemini 4 Argon in AI Studio

Google AI Studio now lets you generate complete native Android apps from plain English prompts. With Gemini 4 Argon, the process handles complex multi-step projects better than earlier models.

You describe the app. The agent writes Kotlin and Jetpack Compose code. You preview it in a browser emulator, install it on a phone, and iterate without leaving the browser.

This guide shows the exact steps from the official Google AI Studio documentation.

What you need before you start

Open aistudio.google.com in a desktop browser and sign in with a Google account. Gemini 4 Argon works on the paid Gemini API tier or with a Google AI Ultra subscription. Free plans do not include Argon.

No Android Studio install is required for the initial build. The browser emulator handles previewing.

Open Build mode and select Android

Click Build in the left navigation. In the platform picker, choose Android. This switches the agent to generate a Kotlin and Jetpack Compose project instead of a web app.

The interface shows a prompt box ready for your description.

Write a clear first prompt

Start with a specific description. Official examples include a daily task tracker with local storage or a recipe app that suggests meals from ingredients.

Try this prompt:

Create a daily habit tracker Android app. Include a list of habits, a calendar view for streaks, and local storage so data persists between sessions. Use a clean Material Design look with Jetpack Compose.

Click Build. The agent generates the full project. This can take several minutes. A browser-based Android emulator opens when the first version is ready.

You can then chat with the agent to change colors, add screens, or fix behavior. The agent updates files across the project.

Preview and iterate in the emulator

The embedded emulator runs the generated app. Interact with buttons and lists as you would on a phone.

If something looks wrong, describe the fix in the chat. For example: Change the primary color to teal and add a floating action button for new habits.

The agent applies the change and refreshes the preview.

Install on a physical Android device

When the preview looks good, click Install. AI Studio uses the browser ADB to push the APK to a USB-connected phone.

Enable Developer Options on the phone by tapping Build Number seven times in Settings > About Phone. Then turn on USB Debugging under Developer Options.

Connect the phone with a data-capable USB cable and follow the on-screen prompts in AI Studio.

Add Firebase for data and AI features

If your app needs a database, sign-in, or Gemini calls, mention Firebase in a follow-up prompt. Example: Add Firebase Firestore so users can sync habits across devices and Firebase Authentication with Google Sign-In.

The agent can provision the Firebase project, generate google-services.json, and write the integration code. For AI features it sets up Firebase AI Logic and App Check so the app calls Gemini without embedding API keys.

Confirm the Firebase setup card when it appears. You can choose an existing Google Cloud project or let it create one.

Publish for testing on Google Play

AI Studio can create a Play Console app record and upload an internal testing build. You need a Google Play developer account. The process packages the AAB and places it on an internal track so you can install updates quickly.

For a full production release you still move to the Play Console for store listing, screenshots, and review.

Tips for better results with Gemini 4 Argon

Argon keeps context across long sessions. Refer to earlier screens when you ask for changes instead of repeating the whole app description.

Break large features into smaller prompts. First get the core list working, then add the calendar, then Firebase.

If the generated code has a bug the preview shows, describe the exact symptom. The agent can read the current files and propose a fix.

For more advanced local development, export the project and open it in Android Studio. You can also connect Android CLI and Antigravity if you prefer an agentic terminal workflow. See our earlier guide on setting up Android CLI for agents.

Watch the official walkthrough

Sources

• Google AI for Developers: Build Android apps in Google AI Studio
• Google AI Studio: Build with Gemini 4 Argon
• Android Developers Blog: Build native Android apps in Google AI Studio
• Firebase docs: Add Firebase to your Google AI Studio app