Receipt Raven

Inspiration

Receipts are easy to lose, especially now that people deal with both printed receipts and digital receipts. That makes expense tracking, warranties, reimbursements, and tax preparation stressful. Most people only think about receipts when they already need them.

We wanted to break that norm by making receipt tracking feel less like boring paperwork and more like a fun habit.

That is how we came up with Receipt Raven: a gamified receipt scanner where users scan receipts, organize them, and slowly win the affection of Reece the Raven, a sarcastic but helpful mascot.

What it does

Receipt Raven lets users scan or upload receipts, then uses OCR and AI to turn messy receipt images into useful structured data.

Users can:

Take a photo of a physical receipt
Upload a screenshot or image of a digital receipt
Extract text using Google Cloud Vision OCR
Parse receipt details such as merchant, date, subtotal, tax, total, and category
Save receipts into a local receipt history
Track total spending and tax-related receipts
Build Reece the Raven’s Affection Meter by scanning more receipts

The fun part is Reece. At low affection, Reece roasts the user in a helpful way. As the user scans more receipts, Reece becomes more affectionate and supportive. We also added voice so Reece can speak after scanning a receipt, making the app feel more alive.

How we built it

We built Receipt Raven as a mobile-first web app inspired by clean iOS-style interfaces. The app has separate flows for recording a new receipt with the phone camera and uploading an existing receipt image.

The core flow is:

The user records or uploads a receipt image.
Google Cloud Vision OCR extracts the raw receipt text.
Gemini parses the OCR text into structured receipt fields.
The user reviews and saves the parsed receipt.
Reece’s Affection Meter increases.
Reece reacts with an affection-based message and voice line.

We kept the design minimal so the app feels simple and approachable on a phone, while still looking good on larger screens.

Challenges we faced

One challenge was receipt parsing accuracy. Receipt formats vary a lot, and OCR text can be messy. For example, totals, taxes, and subtotals may appear on different lines or use different labels. We had to improve the parsing logic so the app prioritizes clear labels like TOTAL, SUB-TOTAL, and tax labels such as HST.

Another challenge was local storage size. Saving full receipt images or large base64 strings can quickly exceed browser storage limits, so we had to keep saved receipt data lightweight and avoid storing large image data.

We also had to balance fun with usefulness. We wanted Reece to be funny and memorable, but still helpful instead of annoying or distracting.

Accomplishments that we are proud of

We are proud that we built a working end-to-end receipt scanning experience in a short hackathon timeframe.

Some highlights:

Physical receipt scanning through the phone camera
Digital receipt image upload
Google Cloud Vision OCR integration
Gemini-powered receipt parsing
Receipt history and spending summary
Affection Meter gamification
Reece the Raven as an animated mascot
Affection-based personality system
Voice feedback after scanning

Most importantly, we turned a boring real-life task into something playful.

What we learned

We learned how to connect multiple AI and cloud services into one user flow. We also learned how important fallback logic is when working with OCR and AI parsing, because real-world receipt data is messy.

On the product side, we learned that a hackathon project becomes much stronger when it has both a useful purpose and a fun emotional hook. Reece helped make the app feel memorable instead of just being another receipt scanner.