CrisisMatch

Inspiration

The news is usually just people talking about problems without doing anything to fix them. We saw big crises like floods and wars and felt like we were just watching a movie from our couches. We wanted to build something that actually tells people what to do. We wanted to turn viewers into helpers by connecting real data to real people.

What it does

CrisisMatch is like a super smart radar for the whole world. It follows ships and planes in real time near disaster zones so you can see how supplies are moving. It uses three different AI brains to argue about how dangerous a situation is. Once the AI decides what is needed it automatically picks the best volunteers from a list to go help out.

How we built it

We used React for the website and Node.js for the server. We hooked up a live feed for ships using WebSockets so the icons move on their own. We used Claude and Gemini as our AI agents. We made sure the app still works even if the internet gets slow by using backup data.

Challenges we ran into

Live data is a total nightmare to work with. Ships move and planes vanish from the map if the connection blips for even a second. Getting three different AI models to talk to each other was like babysitting three toddlers who all think they are the boss. We had to write a lot of code just to make sure they did not time out or crash the site.

Accomplishments that we're proud of

We made the AI agents run at the exact same time so you do not have to wait forever for an answer. We also built a system that hides volunteer names to keep them safe. The live map actually feels alive because it updates every few seconds with real ship positions.

What we learned

We learned that you cannot trust one single AI to have all the answers. Sometimes Gemini is better at news and Claude is better at logic. We also learned that building for a hackathon means things will definitely break. You need a backup plan for every single button and feature.

What's next for CrisisMatch

We want to add a feature where the AI finds brand new crises on the internet by itself. We also want to make the volunteer matching even smarter by using a math formula for scores. The matching logic uses a simple weighted sum to rank people.

$$S = \sum_{i=1}^{3} w_i x_i$$

In this formula S is the final match score while w represents the importance of a specific skill.