In class, we noticed girls casually browsing through clothes on their laptops. We found that the digital shopping experience, whether on mobile or laptop, is overwhelming with generic catalogs and inconsistent experiences. We wanted to create an experience that was easily digestible across a shopper’s favorite brands. We saw an opportunity and created Fleek’s mobile app, with full-bleed images, multiple brands, and a Tinder-like swiping interface.

What it does

Today, we are Tinder for fashion. Our app creates a shopping experience focused on fashion discovery using AI. Simply by swiping, Fleek finds and recommends the latest fashion for your style. You can filter items, save and share favorites, and purchase them.

How we built it

Fleek is a native mobile app, built for iOS using Swift. My co-founder Cyp and I mock up designs using Figma and I subsequently implement them using the standard UIKit framework with Autolayout. My co-founder Kian has built out a Spark streaming backend infrastructure and a content based recommendation model. We pull new items from our affiliate partners on a daily basis and run recommendations hourly. On the app, I pull the clothing items for a specific user with a get request which is hooked up to our backend database.

Challenges we ran into

Our first challenge was figuring out how to get content on the app. We did some research and found out about 3rd party affiliate programs such as CJ and Rakuten. We applied to become affiliate partners with brands suited to our target demographic and learned how to use their APIs to pull clothing items for our app.

I realized that people love to share items with their friends and that our app had an organically viral aspect to it. However, if someone shared an item with a user, a generic link would be shared and it would redirect the user to the clothing item in its respective brand's website. There was no link back to us and we wanted to change that. I dug into universal links and created an experience which opens up Fleek with the appropriate item when it is shared with them.

Accomplishments that I'm proud of

Since releasing our updated app on July 24th, we have a total of 572 downloads. Of these new users, 120 are weekly active users. Our average user swipes 130 times with 50 users having swiped over 1000 times. I'm very proud of these numbers because even though our number of users aren't very high, the usage per user has been great. I'm also very happy that personalized items have been 6 times more likely to be faved than random ones, which means our recommendation model is working great.

I'm simply proud that we were able to put together a fully functional product that could put a smile on someone's face and be useful to them. I'm proud of the attention to detail we put into Fleek's UI. I particularly like the custom view controller animation when tapping on an item from the home page and the heart icon at the bottom right beating when a user swipes right.

What I learned

We are learning so much in the process of building Fleek. Despite having worked on iOS apps for a while, I have picked up a bunch of new technologies I haven't used before. I've learned to use collection view diffable data sources, universal links, context menus, along with many others. I've even expanded outside my comfort zone and done a little bit of web development. I've learned how to host a website on Google Cloud using App Engine and how to write REST API's in Python using Flask.

Apart from the technology, we're continuously learning about our target demographic and how they like to shop. We are constantly asking them questions about their favorite brands or features they'd like on Fleek and gaining a lot of insights in the process.

What's next for Fleek

Fleek has a long way to go. We want to take our product to the next level with a whole lot of new brands and features such as similar items, collections (playlists), search, and a user profile with social and community components. We want Fleek to be people's go to fashion discovery and shopping app.

Our next steps include partnering directly with the brands instead of going through services like Rakuten. That way, we can enable features like in-app checkout which would make buying items on Fleek a lot easier. Moreover, we feel like we can provide fashion brands valuable insights about their clothing items. We envision a future where brands can test the waters with new items before beginning the manufacturing process so they can tailor supply based on what is in demand. This could save them a lot of costs and be good for our environment as there would be less waste.

Alongside that, we want to continue to grow the Fleek community by getting more users, growing our Instagram page, and getting campus ambassadors across universities.

Built With

Share this project: