If you’re interested in translating or adapting this post, pleaseemail us first.
There’s no doubt the FEED app is one of the top three most interesting mobile projects that we’ve recently worked on. Together with the product team, we created and launched an extremely sophisticated B2C mobile product that utilizes cutting-edge technology, including machine learning and social mechanics, from scratch. We’ve poured our hearts into this project, and we believe that it has enormous potential to change the way people communicate online.
The concept and features behind FEED
As part of the global shift towards online communication, Evil Martians created a collaborative space under the creative direction of Heaven Production, the company behind FEED. While most social media outlets are focused on the self, FEED shifts the narrative by focusing on collective conversations.
At its core, the app provides an inclusive and inspiring environment for people to connect across countries, backgrounds, and cultures. Creators select a topic of interest and invite everyone to participate and contribute their point of view. Everything occurs via video, there’s no text: you can post a video to start a conversation, reply with a video, or send your video as a direct message.
Afterward, the app analyses each video and produces a film composed from the best bits of the conversation. This film (or collection of clips) can then be shared on the user’s profile or other social platforms.
Many exciting features allow users robust options to engage with and contribute their own personal videos on their FEED profiles. The advanced video composer has features like pinch-to-zoom, trim and cut functionality, autofocus, a beautifier, color filters, a denoiser, and a built-in music assistant. Taken together, these allow users to produce high-quality video. A walkie-talkie mode lets users record immediate videos with just one tap. The video reply feature replaces text comments. A user profile keeps users’ favorite memories at the tips of their fingers while its log section saves their replies.
The application was released at the pinnacle of the social media video and audio boom, which allowed FEED’s founders to conduct quality market research and effectively hone in on the qualities users really want: asynchronous communication, an elegant and user-friendly design, and free access.
Building the app from scratch
The project’s concept allowed us to adopt cutting-edge tech. We used Swift as our primary language for the iOS app, Swift UI and UIKit for user interfaces, and the latest Ruby on Rails on the backend with GraphQL for the APIs, hosted with Kubernetes on Google Cloud. And many other technologies, including machine learning.
Learn more about how our Martian Design Sprint helps our customers elevate their ideas and grow faster.
Through collaboration with the product team, Evil Martians shaped the general product architecture, its critical features and users’ golden flow, and release management with QA and product analytics. An MVP was required to be released in three to four months, so we kicked things off as we often do—with a targeted design sprint.
Led by FEED’s powerful creative team, we adapted the initial design concept to conform with iOS guidelines, developed a design system, and established the initial versions of the app’s main screens.
On the mobile frontier, we created complex and advanced-level features: a video composer that allows users to record and edit videos, “films” (the aforementioned collections which feature the most popular videos from a certain feed), login, registration, onboarding, navigation, and many others.
Meanwhile, on the backend, we handled essentially everything, including the app’s architecture, all of FEED’s API, statistics databases, search functionality for feeds and users, and the ability to filter and rank search results to help fresh and popular content gain more momentum. We also implemented a lot of optimizations and load relief, like squeezing videos to the last byte for performance, shrinking film compilation time from 30 minutes to a few milliseconds, and experimenting with video size to balance the best resolution with the smallest file size.
We built machine learning tools for the app, both to provide a smoother, smarter, and safer overall user experience, and to support some of its most sophisticated features. We implemented spoken language identification abilities for the videos (including English, Italian, German, French, Russian, and more). We also developed NSFW content recognition, speech-to-text functionality, and multilingual text classification to identify which topics users are talking about, and algorithms to adapt audio quality as needed.
The client’s marketing team, with the help of Evil Martians, ran multiple rounds of near-public user tests with real people—actors, influencers, journalists—on the app’s Beta version. We ran meticulous user tests which hosted 300+ users, saw 1,000+ videos recorded, with 250 minutes of content created, and 9,000+ views on a single feed overall. These allowed the engineering team to optimize the app’s performance and minimize infrastructure costs.
The Martian team recorded their own “feeds” daily to see how the app behaved over time. As we noted in the beginning this was truly a team effort that everyone put their hearts into.
Finally, we prepared the project’s infrastructure and deployed the app to production. As a final touch, we created a presentation for the FEED team with recommendations to strengthen the app’s security and to enable fast scaling.
Launching this app on the App Store was an exciting part of the journey for the whole team—we even filmed this emotional moment.
In April 2021, FEED launched for 35 countries in the AppStore, including the USA, Canada, and most of Europe. The FEED team, along with their partners, have also launched the Android version. In June 2021, the app officially launched at Product Hunt and was endorsed by product hunter Matt Navarra, establishing itself as the #5 Product of the Day.
After the official release, the client’s marketing team created a strategy involving influencers and creators from different backgrounds. They started intense marketing activities to inspire more users to join this innovative and completely new form of communication. FEED’s creative team also developed an informative promo video. For monetization purposes, FEED’s team created a customized advertising account to launch three types of advertising in the application, and Evil Martians helped with the integration of this account on the frontend.
Being involved with developing this creative and innovative social platform has been an amazing experience for the team. We’re beyond excited for everyone to experience this new world of online communication.
If you’re interested in collaborating with us on your product—working from scratch or at any level of development—please reach out to us!