SplitPay AI

Inspiration

Splitting a restaurant bill should be the easy part of a great meal. But when everyone ordered something different, figuring out tax and tip—and making sure nobody pays a few cents too much—can turn into an awkward group chat. I built SplitPay AI to make that part feel a little less like work. Upload a receipt, describe who had what in plain English, and the app works out each person’s share. Friends can then pay through individual PayPal Sandbox checkout links, while the dashboard tracks who has paid.

How I built it

The app uses Google Gemini to read receipt images and understand natural-language instructions. For the actual bill math, I chose not to rely on AI: a Python calculation engine distributes amounts in integer cents, including tax and tip, so the shares add up exactly to the bill total. A FastAPI backend connects receipt processing, bill storage, split calculations, and PayPal Orders v2. The browser interface is built with HTML, CSS, and JavaScript, with AG Grid for reviewing the split and QR codes for sharing checkout links. SQLite stores bills and payment status.

What I learned

I learned how important it is to give AI the right job. Gemini is useful for interpreting messy receipts and everyday language, but financial calculations need predictable, testable rules. Separating those responsibilities made the results easier to verify. I also learned that connecting services is only part of the challenge: payment returns, saved bill state, currency conversion, and clear error messages all matter to the experience.

Built With

• ag-grid
• artificial-intelligence
• bill-splitting
• fastapi
• google-gemini
• javascript
• payments
• paypal
• python
• qr-codes
• receipt-ocr
• sqlite
• web-app