Antigravity with Gemini 3.7 Flash-App for Climate Change Alerts for Grocery Shops

How are you doing?

Summary: Antigravity by Google with Gemini 3.7 Flash is good, intelligent, and quick, but it needs to be told to separate data from code, an essential step for good coding practice.

Let’s see how to make an app or website in minutes with Antigravity Gemini 3.7 Flash.

Here is my experience with it.

My prompt:

“I want you to make a rain app for grocery stores so that stores 
know when rain has to come and inform the buyers that delivery men cant deliver in this time frame.”

I entered the above prompt in Antigravity: I wanted to make it for forest fires and earthquakes as well. But for now, see a shortened version for rain. Until I make it for Climate Change fully.

It asked me several times to accept some permissions to write on system and to install some packages. I said yes in all that.

And within a minute the website was made. It gave me a plan as seen on right panel in the above image.

Then I approved the plan without any change.

It intelligently made a plan all on its own.

Then my website was up and running on: http://localhost:8000/

Here are the alert messages on the website:

You can deploy it too;

Here are the files.

I have uploaded the code on GitHub here:https://github.com/Nidhika1121/AntigravityWithGemini_3_7Flash/tree/main/Code1

Some improvements and future work that are required:

Names are hardcoded. We need to store people’s names, cities, and phone numbers in an Excel file and read them from there. The following server.py file contains hardcoded names, addresses and phone number. Many hardcodes are also present in index.html. So Antigravity by Google with Gemini 3.7 Flash is good and quick, but it needs to be told we need to separate data from code as a good coding practice.

In future work, we plan to add forest fires, earthquakes, and cloudbursts to this app and remove hardcodes from the code. All with Agentic AI. With data pulled from a database or Excel files, not hardcoded data. Hardcoding data is not of much good.

Thanks for reading.

Subscribe for updates.