Inspiration
We wanted to use a fast, affordable classifier to handle financial review work that people find tedious.
What it does
Fishy scans financial records and points internal teams toward suspicious transactions worth reviewing.
How we built it
We built a classification system that checks each record for risk signals, then combines the results into one score.
Challenges we ran into
The hardest part was deciding how to present the finished analysis so the results were useful and easy to understand.
Accomplishments that we're proud of
We’re proud of the classification step and the scoring algorithm behind it. It can sort through a lot of information without requiring much time or money.
What we learned
We learned how classifiers work and how to turn their probabilities into results people can act on.
What's next for Fishy
We want to improve the classifier, support larger analyses, and make the results even clearer for small internal teams.
Built With
- bun
- mammoth.js
- next.js
- ogl
- react
- sheetjs
- tailwind-css
- typesafe-ai
- typescript
- unpdf
Log in or sign up for Devpost to join the conversation.