BuddyBillder – Bill Splitting, Made Effortless

🚴‍♂️ The Inspiration

We often go cycling in and around Amsterdam—sometimes just the usual routes, sometimes hilariously random places. After a good 2–3 hour ride, we usually stop by a café to refuel. One person pays the full bill, and then the awkward part begins: figuring out who owes what.

People forget to pay. Nobody remembers exactly what they ordered. Calculating the split takes effort, and it ruins the post-ride vibe. We thought—why not make this frictionless?

💡 The Idea

What if you could just take a photo of your food and let AI figure out what you ate? Then—bam—pay your share directly via bunq. No spreadsheets, no memory games. Just tap and done.

So we set out to build it during the bunq hackathon. And honestly, bunq is the perfect playground for this.

👥 The Team

We’re a team of three:

Two developers from Booking.com
One Computer Science student from TU Delft

🛠️ How We Built It

We started by designing the user journey in Figma, mapping out the experience from receipt upload to payment. Then we split responsibilities:

Backend: handles receipt parsing, payment logic, and bunq API integration
Frontend: displays the receipt, supports food photo uploads, and user interaction
AI Module: recognizes receipt items and matches them with food photos

Finally, we integrated all components—and it worked! 🎉

🧠 What We Learned

OCR for receipts is trickier than it looks
Matching food images to menu items is both fun and hard
Good planning and clear task ownership helped us deliver fast
And… bunq’s API is a joy to work with

😅 Challenges

Getting accurate results from OCR across different receipt formats
Making AI accurately match food photos with receipt lines
Aligning our work across different tech stacks and stitching it all together under time pressure