Subletly

Inspiration

As university students in co-op programs, we relocate to a new city every academic term for internships. Finding short-term housing is stressful, especially when decisions often need to be made before ever setting foot in the city. We wanted a tool that could quickly show what kinds of sublets would realistically fit our budget, lifestyle, and transportation needs, without endless searching or guesswork.

What it does

Subletly is an AI-powered web platform that helps users discover sublet options tailored to their preferences. Users input their budget, city, desired amenities (such as parking, pet-friendly spaces, and WiFi), and accessibility preferences like walkability, transit access, or car-friendliness. Subletly then generates realistic sublet listings using AI, assigns each one a match percentage based on how well it fits the user’s criteria, and clearly displays the amenities included so users can easily compare options.

How we built it

We built Subletly as a full-stack web application. The frontend was developed using Next.js and React, with Tailwind CSS, HTML5, and CSS3 to create a clean and responsive user interface. The backend was built using Python and Flask, which handles user inputs, filtering logic, and communication with the AI model. We integrated the Google Gemini API to generate realistic sublet listings based on user preferences. JavaScript and Node.js were used to support frontend logic and API interactions, and we implemented a matching algorithm that scores each listing to produce an intuitive match percentage.

Challenges we ran into

One major challenge was ensuring that AI-generated sublets were realistic, consistent, and aligned with user constraints like budget and amenities. Designing prompts that reliably produced structured and useful results took significant iteration. Another challenge was defining how different preferences should be weighted when calculating match percentages so that results felt intuitive rather than arbitrary. Integrating multiple technologies across the frontend and backend within a limited hackathon timeframe also required careful coordination.

Accomplishments that we're proud of

We’re proud of building a complete, end-to-end product that feels practical and easy to use. Successfully integrating Google Gemini in a meaningful way, rather than as a novelty, was a major accomplishment. We’re also proud of the clear match percentage system and amenity breakdowns, which help users immediately understand why a sublet is or isn’t a good fit. Additionally, this project was a huge learning experience for one of our team members, as it was his first hackathon and gave him hands-on exposure to web development, full-stack collaboration, and the fast-paced hackathon environment.

What we learned

Through this project, we learned how to effectively integrate generative AI into a real application workflow, from prompt engineering to validating outputs. We gained hands-on experience connecting a React and Next.js frontend to a Python Flask backend, and learned how to design scoring systems that translate subjective user preferences into clear metrics. We also learned how to prioritize features and iterate quickly under hackathon constraints.

What's next for Subletly

Next, we want to integrate real sublet data from existing housing platforms and allow users to save and compare listings across sessions. We also plan to refine the matching algorithm using user feedback to improve personalization. Long-term, we see Subletly expanding beyond students to support interns, digital nomads, and anyone relocating to a new city for short-term living.