Google just released their incredible Nano Banana model and it’s been wild these past few days.
The major game changer and my favorite feature is the stunning image editing ability.
Which is why many people have been calling it the Photoshop killer.
And for us devs this just opened up a massive world of opportunity to build the most amazing image-focused apps.
I’m gonna show you how we can so easily create such an app in just like 5 minutes with the awesome power of coding agents.
Before deciding what you want to build you have the Google AI Studio to help you play around with Nano Banana — and all the other Google image models.
With the Studio you can easily check how good the model is for your use case.
Test and refine prompts until they give you exactly what you’re looking for
Unleash your creativity without limits.
So let’s say we’re creating an app to bring a grayscale image to life with color.
Users upload any image in grayscale and they get a beautiful colorized image — giggling and all excited to make it their wallpaper and tattoo it on their back.
Before image models like Nano Banana this would have a lot less straightforward — using specialized Python libraries to process the image through multiple stages and stuff.
But now all it takes is just 3 super simple words of prompting.
Let’s even say 1 word cause i bet it would work with just “colorize”.
Yes:
So just one word is all it takes now with Nano Banana.
So imagine how easy the app is going to be to make — the core engine is already fully taken care of by AI.
With vibe coding and AI agents the rest is just so easy.
We don’t even need any framework like React or god forbid Next.js — just basic HTML, CSS, and JS will do — especially since we won’t be writing a single atom of the code ourselves.
First we just use the “Get code” button in the Studio to get the JavaScript code version — let’s us use our ridiculously simple prompt in a real server.
This is another brilliant perk of using Google AI Studio — easily get the boilerplate code to use any prompt for any AI model.
Creating the actual server is just so easy now:
Look I just said:
“create an express server with a /processImage route to colorize an uploaded grayscale image using this code:”
And then I pasted the code from the Studio.
And that was that.
You see first of all the Cascade agent created a clear todo list so it knows exactly what it’s doing — yeah sadly we humans don’t have a monopoly on todo lists anymore…
And then BOOM — every single task with its sub-tasks and sub-sub-tasks — thoroughly conquered and vanquished in a matter of minutes.
At the end of the day:
Every single file and code here came from the coding agent.
Wow the agent even added code to remind me to use my GEMINI_API_KEY — something I actually forgot.
Getting an API key is super easy — just use Get API key -> Create API key in the Studio.
Even more incredible — it generated the entire client-side code for me.
I only asked and expected just the server side server with routes and all — but it went the extra mile and did everything.
All done:
So I uploaded the grayscale city view from earlier:
And in a few seconds:
In just 5 minutes — or maybe even less — I created this incredible colorizing app from absolutely nowhere.
This is the insane power AI gives you.