Inspiration
Every semester, students throw away furniture, electronics, school supplies, and dorm essentials that still have value, while other students struggle to afford those same items. We wanted to build something that strengthened the SFSU community by making giving just as easy as receiving, with our pain points being targeted the ease of creating listings, lower the barrier of entry, easing the transportation costs of picking products up which our moat at e-Broke compared to markets like Facebook market, reputability among buyers and sellers as it is exclusive to college students only. Our goal was to reduce campus waste, promote sustainability, and use artificial intelligence to lower the barrier for students who want to participate in a trusted local exchange.
What it does
e-Broke is an AI-powered, student-only marketplace where verified SFSU students can give away and claim free items. Users authenticate with their SFSU email, browse items using semantic search, message other students, and create listings with AI-generated titles, descriptions, and categories.
Its equity-focused wish-list system identifies students already looking for a newly listed item and alerts them first, giving donors the opportunity to reach out directly to people who may need it most. This makes exchanges faster, more intentional, and more accessible while keeping the platform local and completely free.
How we built it
We built e-Broke using React, TypeScript, and Vite for the frontend, with a Node.js backend connected to PostgreSQL and pgvector for semantic search. AI assists users with listing creation, item discovery, wish-list matching, and future moderation capabilities. The platform is hosted on DigitalOcean and designed around a secure, campus-only experience through SFSU email verification.
Challenges we ran into
One of our biggest challenges was learning how to use DigitalOcean’s vector database capabilities for semantic search, similarity matching, and wish-list prioritization. We also explored agentic AI workflows for automated content moderation, including how specialized AI agents could identify spam, inappropriate listings, and bad actors while minimizing false positives. Balancing automation, fairness, privacy, and user trust was one of the most technically challenging parts of the project.
Accomplishments that we're proud of
We are proud of building a complete campus-focused platform that solves a real problem students experience every semester. We successfully combined AI-assisted listing creation, semantic search, verified SFSU authentication, messaging, and equity-focused wish-list alerts into one experience. Most importantly, we built a platform that helps useful items reach students who are actively looking for them instead of ending up in a landfill.
What we learned
This project taught us how to integrate AI beyond a traditional chatbot by applying it directly to useful workflows. We learned about vector embeddings, semantic search, priority matching, DigitalOcean’s AI infrastructure, AI-assisted user experiences, and how agentic AI can support trust and safety. We also learned how important transparency and fairness are when AI influences which users receive alerts first.
What's next for e-Broke
Our next steps are to improve the wish-list prioritization system, add user-controlled urgency and need preferences, and allow donors to reach out directly to matched students before publishing an item broadly. We also plan to deploy AI-powered moderation agents, strengthen personalized recommendations, improve accessibility, and introduce environmental-impact analytics that estimate how much waste is diverted from landfills.
Looking further ahead, we hope to bring e-Broke to other universities while keeping every campus community private, verified, equitable, and locally connected.
Log in or sign up for Devpost to join the conversation.