Inspiration

Over February Break, Scott and Ezra met an unhoused man from Cameroon who only spoke French. After buying him dinner and a toothbrush, we used Google Translate to determine his needs. However, we had to call four different homeless shelters to find a place for the man to stay.

What it does

When an unhoused individual reaches out to our ShelterFlow chatbot, our app calls nearby homeless shelters before returning a recommendation of next steps for the user.

How we built it

We built ShelterFlow using Claude Code for development, Expo for the frontend, Express for the backend, OpenAI for the chatbot, and Google Places API for geographic searching.

Challenges we ran into

We ran into many challenges throughout: Chelsea helped Scott resolve some merge conflicts that arose early in the development process. Testing the product was difficult since we needed to replace all of the nonprofit phone numbers with phone numbers from our group members.

Accomplishments that we're proud of

We're super proud of integrating so many different tech tools together into a working mobile application!

What we learned

We learned that it's important to plan out the entire user flow ahead of time to ensure we understand all of the technical qualifications required to ultimately maximize our social impact.

What's next for ShelterFlow

We will expand out the scope of the chatbot to support more than just food and shelter, helping to reduce the total system burden from translators.

Share this project:

Updates