transaction categorizer

webpage

Inspiration

Managing personal finances is messy. Most apps categorize transactions statically, which rarely matches how I think about my spending. I wanted a system that adapts dynamically to my own rules.

What it does

Smart Transaction Categorizer lets users create custom rules to automatically re-label transactions. For example: “If it’s Saturday, at Walmart, under $80 → Hardware.” It instantly organizes your spending exactly the way you want.

How we built it

I used Bun for a lightweight backend, local storage to store transactions and rules, and React + Tailwind for a responsive frontend. The backend applies rules to transactions dynamically, and the UI updates instantly when rules change. I am using Gemini for natural-language-to-DSL translation. I am using OpenAI's whisper model for audio transcription.

Challenges we ran into

Designing a flexible rule engine that supports multiple conditions (merchant, day, amount, account).
Handling rule conflicts and ensuring transactions update correctly in real time.
OpenAI and Gemini integration

Accomplishments that we're proud of

Built a fully working dynamic categorization system
Users can create, edit, and delete rules, and see transaction categories update instantly.
Developed a rule engine that can handle AND/OR conditions and multiple criteria.