Inspiration
The secondhand clothing market is booming, driven by a desire for sustainability and unique fashion. However, the market is highly fragmented. Shoppers must manually search across dozens of different websites and apps (like Poshmark, Depop, eBay, and ThredUp) to find a specific item. This process is time-consuming and inefficient. Our goal was to promote sustainable fashion by removing this friction, creating a single, intelligent platform that aggregates listings and makes finding secondhand goods as easy as a single search.
What it does
Thrift Finder is a mobile application that acts as an intelligent search aggregator for secondhand clothing.
Natural Language Search: A user inputs a descriptive query, such as "vintage 80s blue windbreaker" or "brown leather combat boots."
AI-Powered Aggregation: The request is sent to our backend, which uses an AI model (leveraging the Claude API) to interpret the query. The AI then intelligently searches across multiple major secondhand e-commerce platforms, parsing the results.
Unified Results: The app displays a single, consolidated feed of results from all sources. Each listing is standardized and includes a description, price, and source.
Save & Purchase: Users can "save" items to a personal wishlist within the app or tap a link to be taken directly to the seller's page to make a purchase.
How we built it
Frontend (Client): We used React Native to build a cross-platform mobile application, focusing initially on Android. This allowed us to leverage our JavaScript knowledge to create a native look and feel with a single codebase.
Backend (Server): The backend was built with Node.js and Express.js. This server acts as the middle-layer, managing user requests and communicating with third-party APIs.
AI & Data: The core search functionality is powered by the Claude AI API. We send the user's natural language query to the API and prompt it to act as an expert shopper, translating the query into effective search terms for various sites and returning structured data (in JSON format) from the web.
Challenges we ran into
While setting up the React Native environment and Android Studio emulators was an initial hurdle, our most significant technical challenge was data normalization. Each e-commerce site structures its data differently. An item's size, condition, or brand might be listed in completely different formats. Our backend had to efficiently parse this inconsistent data, deduplicate listings, and standardize it all into a clean, uniform format to be displayed on the frontend. Integrating the AI to return consistently structured JSON data also required significant prompt engineering and error handling.
Accomplishments that we're proud of
We are most proud of building a fully functional, end-to-end application that solves a real-world problem. Successfully integrating a powerful AI API to handle the core logic of our app—the complex search and aggregation—was a major achievement for our team. We managed to create a seamless user experience that hides the complex backend processes from the user.
What we learned
This project was a deep dive into full-stack mobile development. We gained significant hands-on experience in a lot of new things!
What's next for Thrift Finder
We have a clear roadmap for expanding the app's capabilities:
Advanced Filtering: Adding robust filters for size, brand, price range, and condition.
Price Alerts: Allowing users to save a search query and receive a notification when a matching item below a certain price is listed.
Image-Based Search: Implementing a feature where users can upload a photo of a clothing item, and the AI will search for visually similar items.
UI/UX Refinement: A full redesign of the UI to create a sleek, modern, and intuitive user experience.
Log in or sign up for Devpost to join the conversation.