Mosaic: Youth Homelessness Project

Inspiration

One of our team members has a friend who experienced homelessness as a teen. She was living in her car and spent many months before she even realized that she was "homeless". We were inspired to make a digital assistant that would advocate for her to get the resources she deserved.

What it does

Mosaic can empathetically chat with you to understand your unique needs and guide you step-on-step on how to get access to basic needs resources. It can also connect you with peers to help you form your own support networks.

How we built it

We built Mosaic using Next.js, Vercel, and the Vercel AI SDK with the OpenAI API to create empathetic, real-time conversations. Our data layer uses Drizzle ORM with Neon Postgres for secure, scalable storage.

Challenges we ran into & What's next for Mosaic

Currently, people are required to visit a shelter or resource center in-person in order to complete their registration and get access to resources. We hope that Mosaic can open the door to connecting directly with local coordinated entries to make this process fully online so that people don't have to waste time checking each location, not knowing if they are even eligible for their services.

Accomplishments that we're proud of

We are proud that we are able to create a project that is empathetic and responds to a person's unique needs. Being on your own, without a secure place to live, can be as complex as it is stressful. Having a digital advocate that meets you where you are is Mosaic's mission.

What we learned

The process of getting basic needs resources as a young person, is a complex and scary process. There are strict eligibility requirements that can exclude you from accessing a resource based on how you answer a question. Without an advocate to guide you, it can feel like you have nowhere to go. That's why we started Mosaic, a virtual advocate to help you maximize your access to benefits and connect you with a network of support while keeping you safe.

Built With

• claude
• nextjs