← Back to Blog

How to Build Your First App in Google AI Studio

You don't need to be a coder to build functional apps with AI. Google AI Studio lets you describe your idea in plain English and generates a working application—complete with the ability to iterate, troubleshoot, and deploy. This guide walks through building a minimalist data visualization app from scratch, including real troubleshooting when things don't work perfectly.

Key Takeaways

  • No coding required: Describe what you want in plain English and Google AI Studio builds it
  • Iterate one change at a time: Make single modifications to understand what works
  • Use ChatGPT for troubleshooting: When Gemini struggles, copy the code to ChatGPT for help
  • Restore points save you: Always have the ability to roll back to a working version
  • Start simple: Don't try to build Facebook overnight—master small apps first
  • Test extensively: Every new data input may reveal edge cases to fix
  • Getting Started

  • Go to aistudio.google.com/apps
  • You'll see "Build your ideas with Gemini"
  • Describe your idea in the prompt field
  • Example prompt used: > "Create an app where I can input text on a singular data point, and the app based on that data point, will create a visualization of that data in the style of a minimalist, dark mode circular graph."

    Click "Build" and wait less than a minute. The AI thinks, codes, and delivers a working prototype.

    The Build Process: Real-World Walkthrough

    Initial Output

    First test: "Only 1% of users on LinkedIn create content weekly."

    The app understood the request and generated a circular "pulse visualization." Success on the first try means you can work with this foundation rather than starting over.

    Testing Multiple Data Points

    Before requesting changes, test with various inputs:
  • "0.004% of business owners go live consistently"
  • Various percentage formats
  • Different content topics
  • This reveals patterns and issues before you start modifying.

    Making Iterative Changes

    Rule #1: One change at a time.

    Why? If you ask for four things and the app breaks, you won't know which request caused the problem.

    Change 1: Remove unwanted header > "Get rid of the header that sits above the percentage number. For example, in the recent output, it said 'Business live consistency.'"

    Change 2: Limit text length > "Limit the number of lines for the small text explanation that sits under the big number to three lines max. Make the font slightly bigger."

    After each change, test again with your data points.

    When Things Break: The Troubleshooting Loop

    The app generated: "Only a very small fraction of LinkedIn users actively produce original weak..."

    The word "content" got cut off. Time to troubleshoot.

    Method 1: Tell Gemini Directly

    > "Please improve the logic because this sentence doesn't make sense. It is missing the word 'content' which makes the output unprofessional."

    Sometimes this works. Sometimes it doesn't.

    Method 2: Use ChatGPT (The Power Move)

    When Gemini keeps failing:
    • Click "Code" in Google AI Studio
    • Copy the relevant code (e.g., `gemini-service.ts`)
    • Take a screenshot of the broken output
    • Give both to ChatGPT with your problem description

    ChatGPT will often provide better code or instructions. But here's the key insight:

    When ChatGPT says "go to line 47 and change X": > "I'm not a coder. I need you to rewrite my Gemini service.ts from the top so that I could just copy and paste."

    This gets you complete, working code rather than confusing partial instructions.

    Method 3: Use the Restore Button

    Every change creates a restore point. If ChatGPT's code breaks everything, simply restore to the previous working version and try a different approach.

    Solving Edge Cases

    Problem: 0.0004% displayed as just "0%"

    The fix: > "When a number or data point is something like 0.0004%, do not round the number to zero. Put the exact number."

    Some edge cases aren't worth solving. Ask yourself: "How often will I actually use a number this small?" If rarely, maybe skip it and move on.

    Adding Features

    Once your core functionality works, add capabilities:

    Color Theme Picker: Google AI Studio suggests features automatically. Click "Add color theme picker" and it generates the prompt for you.

    Logo Upload with Brand Colors: > "Add the ability to also upload a logo for the app to identify the colors to leverage within the output."

    Download Functionality: > "Add the ability to download the image."

    Better UX: > "When the image is done being produced, make the save as image button underneath it start to glow slowly so that users know what to do next."

    The Complete Feature Build

    By the end of the session, the app included:
  • Data point input field
  • Circular graph visualization
  • Color theme selection (or auto-detect from logo)
  • Logo upload
  • Properly constrained text descriptions
  • Pulsating download button
  • Image export
  • Total time: About an hour of back-and-forth iteration.

    Key Lessons for Non-Coders

    1. Start Stupidly Simple

    Don't build a social network. Build a single-purpose tool you'll actually use.

    2. Test Before Changing

    Input 3-4 different data points before requesting modifications.

    3. One Change at a Time

    Isolate variables so you can identify what breaks things.

    4. ChatGPT Is Your Backup

    When the native AI struggles, export code to another LLM for help.

    5. Always Have Restore Points

    Never be afraid to experiment because you can always go back.

    6. Rewrite, Don't Patch

    Ask for complete file rewrites instead of partial instructions.

    7. Make It Useful

    If you won't use the app after building it, you're just playing. Build something you need.

    What You're Really Learning

    The app itself is just the outcome. The real skills you're developing:
  • How to prompt effectively
  • How to troubleshoot AI outputs
  • When to switch between AI tools
  • How to iterate toward a working solution
  • Patience with the back-and-forth process
  • "You are learning a lot of the little skills along the way, which is why I said start with something simple."

    Deployment Options

    When your app is ready:
  • Deploy: Publish the app with a shareable link
  • Share: Copy the app and set advanced sharing permissions
  • Export: Download for use elsewhere
  • What's Next

    Ideas for extending this kind of app:
  • Add multiple visualization styles (7+ options to choose from)
  • Include an explainer section for why the data matters
  • Create wide-format images for social media
  • Auto-generate insights for business owners
  • The foundation you build with simple apps prepares you for bigger projects. Master the workflow here, and you'll be ready for anything.

    ---

    Ready to build your first AI-powered app?

    📺 Watch the Complete Tutorial

    Start building at aistudio.google.com/apps

    For more tutorials on AI tools, app building, and business growth, subscribe to Shanee Moret's YouTube channel.

    Want More Insights?

    Subscribe to the YouTube channel for weekly videos on AI, LinkedIn, and growing your business.

    Subscribe on YouTube →