roomer

In a $50B housing market where 32% of adults share homes, finding the right roommates is difficult. Roomer, our AI agent, finds roomies easily with interactive natural language and smart algorithms.

Comment

Inspiration

When finding roommates or shared housing, there's no structured way to find roommates. Each member on our team has lived with roommates and experienced the same issue. With roomer, we enable users to efficiently find shared housing from multiple platforms with their wants and needs.

What it does

Roomer is an AI agent that takes in natural language from the user and matches them with roommate listings that fit their input context.

How we built it

We built roomer using popular frameworks and services like MCP, integrated development environments, and ai assistants.

Challenges we ran into

One challenges we ran into was getting our separate components integrated. We delegated frontend and backend separately so combining these parts took some patience and hard thinking.

Accomplishments that we're proud of

We are most proud of building a dynamic AI agent that utilizes MCP and makes difficult housing market information more accessible.

What we learned

We learned how to utilize the model context protocol and integrate multiple tools and resources.

What's next for roomer

We hope to scale roomer and help as many people we can in finding their ideal roommates.

List of every sponsor tool used:

  • Weight & Biases was used to analyze the tools called from our orchestration agent and determined where we could increase performance and decrease latency
  • Exa was used to perform web crawling and return our location scores per property

List of every protocol used:

  • Model Context Protocol was used to facilitate the connection between our tools including scraping and web search, and resources including our Supabase database

Built With

Share this project:

Updates