A few days ago, I had a simple thought:
“I know my groceries are getting more expensive… but what actually got more expensive?”
I definitely wasn't going to dig through months of receipts to find out.
So I built PriceBack.
It turns your old shopping receipts into a personal price history, so you can see what changed, what got more expensive, and even what quietly got smaller.
And I built the first working version in less than 12 hours with Google AI Studio.
How it works
The flow is pretty simple:
1. Upload a receipt
Take a photo or upload an old receipt. Old receipts are actually useful because they give PriceBack something historical to compare against.
2. Gemini reads it
It extracts the products, prices, quantities, brands, and package sizes.
3. You check the result
AI isn't always right, so nothing is blindly saved. You can edit the extracted products before confirming them.
4. PriceBack remembers
After a few receipts, you start getting an actual price history:
Eggs
March $4.19
June $4.55
September $4.89
↑ 16.7%
My favorite feature: Shrinkflation Detective
Price increases aren't always obvious.
Sometimes this:
8 oz — $4.49
quietly becomes:
7 oz — $4.49
Same price. Less product.
PriceBack compares package sizes and unit prices to catch those changes.
Same price. Less product. Caught.
You can also ask your receipts with Gemini AI
I added Ask PriceBack, where you can ask things like:
“What got more expensive?”
“Why did I spend more this month?”
“Where did I buy eggs cheapest?”
“Did anything get smaller?”
Instead of being a generic chatbot, the answers come from the shopping history you've added to your account.
Built with AI, but I still control the code
This was one of my favorite things about building with Google AI Studio.
I could get from idea → working application incredibly quickly, but I'm not stuck inside a no-code editor.
I can open the project, edit the code manually, change components, debug things, refactor it, and continue building normally.
AI helped me move much faster, but I still have control over what I'm building.
The app uses Gemini + Firebase Authentication + Firestore, with each user's receipts and price history kept separate.
Less than 12 hours later...
I went from:
“I wish I remembered how much this used to cost.”
to an app that can actually answer that question.
There's still plenty I want to improve, but that's also why I like having access to the actual code.
You can try the demo here:
Top comments (0)