Inspiration
Like most teens, I got my dressing sense from TikTok, but couldn’t afford the trendy influencer outfits. In an attempt to thrift, I spent 4 hours finding one jacket that fit my style.
Here’s what people don’t get about fast fashion: there’s a reason it exists. Teens don’t care about long-lasting neutral timeless fashion. We want cheap, fun, and bold designs (which naturally go out of style fast). The market’s response to this is unsustainable and wasteful (a la Zara and Shein). Thrifting is inconvenient and not tailored to different people’s tastes.
The $100 billion second-hand clothing industry has remained painfully stagnant. We knew we could re-imagine e-commerce and build something better.
Environmental Impact
According to US NIST, 85% of used clothes in the US head straight to landfill or incinerators. Fashion is responsible for 10% of global carbon emissions, more than all international flights and maritime shipping combined, according to the World Bank.
Fast fashion is destroying our planet, but people like it too much. We solve this problem by giving users an incentive to be sustainable; everyone wins.
What it does
On our platform, creators can post TikTok-style videos showcasing outfits they want to sell. Consumers can swipe through their video feed, tap on clothes they want to buy, and set up an exchange. A lot of our users will be both buyers and sellers. Naturally, like Craigslist, we’re meant for users living in the same area/college campus/city. Like TikTok, users can discuss trends, mix and match outfits, and share finds with the community. Goodwill and popular thrift pop-ups already exist, we digitize this.
Teens buy clothes based on “aesthetic”. Finding clothes you like at thrift stores is tedious. Our hybrid recommendation engine uses a novel AI model to recommend fashion based on watch time, likes, comments, common interests, past purchases, and more.
How we built it
The SecondSwipe app was made using ReactJS, NextJS, and Firebase. We use the real-time database, Firestore, and Cloud Storage as backends for creator and consumer data. SecondSwipe uses Open AI’s API (text-davinci-003) to suggest a suitable selling price based on the product’s original price, condition, and market standard.
It also uses Checkbook’s API to enable secure end-to-end payments with the tap of a button. What sets SecondSwipe apart, is its novel AI-powered recommendation engine. This is a hybrid model based on deep neural collaborative filtering and knowledge graphs. This model was trained and tested in Python using Keras and connected to the platform using Flask.
Challenges we ran into
SecondSwipe is not only an e-commerce app but also a video-sharing platform. This made it challenging to structure the recommendation algorithm that incorporated features like watch time, likes, user budget, past purchases, etc. all into a single model.
Time: Developing, debugging, and deploying a full-stack working application in less than 36 hours was definitely not an easy task.
What we learned
- How to integrate a secure payment system for the first time
- How to efficiently enable video sharing in React
- Thinking from a sustainability viewpoint in a field where it is not the norm
- How to handle a live YC interview, building products with a user-first mindset
- Surviving on Taco Bell and no sleep
What's next for SecondSwipe
- Advanced computer vision search using Google's Vision API and Web Detection
- Grow user base by going local: targeting communities like college campuses where the population is young and exchanges are convenient
- Developing a business model: monetizing by taking a small percentage from transactions, advertisements (especially by mid-size clothing lines), company collaborations, and subscription services.
- Re-imagining reviews and developing a similar product for fashion companies to showcase on their websites.
Log in or sign up for Devpost to join the conversation.