Car-Fish

Home Page
Question presented by the app
Prompted Input Question

Inspiration

The inspiration for the project clearly comes from modern dating apps like Tinder, Bumble, and Hinge. These apps use swiping as a way to quickly sort through a decently easy question: "Do you like this person?". We took this idea of quickly answering questions and applied it to some of the boring questions on an insurance claim form.

What it does

It takes a list of questions and pre-given information about the claimant to create a "swiping stage". You'll swipe through the set of questions, and when prompted, you will enter some extra details that might be needed. We've got a profile system that can separate multiple claims to one user, but also allow another user to make their own claims. This profile data will autofill part of the form for you. To help further streamline the claiming process, we've implemented a dependency graph in order to not ask redundant data.

e.g., If we ask if the accident happened in a car park, it won't ask you any more questions about the location of the incident; instead, it will auto-fill those questions.

How we built it

We created this in React + Vite with mobile compatibility in mind. We found that the benefits of React Native didn't outweigh the time it would take to learn the framework. Other than that, we've set up a Flask API to allow storage of information about the claims.

Challenges we ran into

The biggest challenge we ran into was the idea generation itself. We wanted to challenge ourselves with something more technical, like AI prediction models to estimate damage cost or having an all-in-one solution for fraud detection, but after much deliberation and discussion, none of those ideas solved the track's challenge. So we shifted, focusing primarily as UX designers rather than trying to reinvent the wheel.

Accomplishments that we're proud of

Neither member of our team is well-versed in frontend development, so I'm really proud of how nicely the project works. We spent considerable time finding the exact pain points users have when filing a claim, which led to our dependency graph idea and the idea of speeding up the process with swiping. This is the first project where I feel that the User Experience and Interaction is the selling point, which is quite foreign to both of us.

What we learned

Tailwind CSS does not flow well with our kind of development. Both of us were working primarily on the front-end lead to multiple merge conflicts throughout the weekend. Tailwind just made the process way worse. Having a global CSS file or separate CSS files would have been fantastic. We learned how to add swipe gestures and how to distinguish between mobile and web views. We learned the very basics of Flask after a year of using FastAPI, and it seems simpler and easier to work with.

What's next for Car-Fish

It would be fantastic if we could hook up an AI model to figure out the dependency graphs on behalf of the claims company, and then dynamically modify the order to prioritise questions that are likely to cut down on total questions asked.