// Case study · Fintech & Dining
A bill-splitting app that reads the receipt and requests each friend's share via Venmo
DIVIT turns a restaurant receipt into a split bill. Its proprietary vision-parsing algorithm reads the receipt, friends are dragged onto the items they ordered, and the app requests each share through Venmo.
- Client
- DIVIT
- Platforms
- iOS · Android
- Industry
- Hospitality
The project
The challenge
Splitting a group dinner is a small, universal hassle. One person pays, then someone has to read the receipt, work out who ordered what, divide the tax, and chase everyone for their share.
DIVIT wanted an app that does that work: scan the receipt, assign each item to the right friend, and request the money. The hard part is the receipt itself. Layouts vary from restaurant to restaurant, and the app has to find the tax, subtotal, total, and every item and price without anyone retyping them.
What we built
We built DIVIT as an iOS and Android app around a proprietary vision-parsing algorithm. It pulls the text from a photo of the receipt and determines the tax, subtotal, total, and each line item with its price.
Once the receipt is parsed, the user drags and drops friends onto the items they ordered. DIVIT works out what each person owes and requests the money through Venmo.
Drag and drop keeps the split fast at the table. Requesting payment through Venmo means friends pay with an app they likely already use, and DIVIT stays out of moving money itself.
The result
- Users scan a receipt instead of typing it in, and the app finds the tax, subtotal, total, and line items.
- Friends are assigned to items by drag and drop, and each share is requested through Venmo.
- The app was built for both iOS and Android.
Built with
- Proprietary vision-parsing algorithm
- Venmo payment requests
Where AI takes it next
How we'd extend this product with today's AI — the same thinking we bring to every client roadmap.
- 01
Any-receipt parsing: a multimodal model (one that reads images and text together) takes the first pass, so faded, crumpled, handwritten, and foreign-language receipts work too, and the line items are checked against the printed subtotal before any request goes out.
- 02
Natural-language splitting: type or say "Sam and I split the nachos, Priya had two margaritas" and the app assigns the items.
- 03
Learned group habits: if a group always splits appetizers evenly or covers the birthday person, DIVIT suggests that split the next time they scan a receipt together.
- 04
Trip mode: an agent that collects every receipt from a weekend away, categorizes them, and nets out who owes whom so the group settles up with the fewest Venmo requests.
Services on this project
What we brought to the table.
Next step
Have a product or process like this?
Tell us what you're trying to build or fix. We'll come back with an approach, a rough timeline and the AI opportunities hiding in it.
More work
Keep exploring
One home for an independent hip-hop artist's music, fans, and merch
Hip-hop artist Justin Starling wanted to pull his fans from scattered platforms into one place. We built his website, a fan text-alert system, and a Shopify merch store, and handled his SEO and Facebook Pixel setup.
One platform for car enthusiasts to meet up, learn, and trade parts
Grand Car Community set out to bring car enthusiasts together. We built its landing page, mobile app, and web application, giving members one place to host local car meets, learn in forums, and trade parts.
A bold sports-betting recommendation platform, built to scale beyond its beta
Sure Pick Sports wanted a flashy, edgy website for its prototype sports-betting recommendation system. We designed and developed the platform with scalability in mind while it was in beta.