The full case study
AI-featured social media app, step by step
On this page · 8 sections
About the business
The story of our cooperation with the client started in 2022, when he came to us with an ambitious idea to revolutionize the current social news model and build a unique AI-powered app.
Our client noticed 2 major problem with the existing news apps:
- they provided repetitive news that came from untrusted sources,
- they were poorly organized in terms of news delivery,
- the news feed in apps always looked messy.
So our client wanted to build an app that will be a go-to solution where users can easily find news, exchange opinions, and enjoy the feed created by AI algorithms and adapted to their interests.
Our client already had a team of marketing, SaaS and e-commerce professionals but they all lacked tech expertise to turn their idea into reality. So they came to Artkai with first drafts of the app visuals, and GTM strategy.
We liked the idea behind the future solution and decided to help this team. We prepared the project plan and roadmap, selected a suitable tech stack for native iOS and Android apps, and began working on the design.
Our goal was to prepare a solid beta version of an AI-powered app to help our client raise further funding. It took us 7 months to build a core functionality and perform a successful beta launch. The following step was further development and public launch which also went smoothly.
Project tasks
- Prepare a detailed project plan with requirements and roadmap to align expectations with tech capabilities and understand scope of work.
- Select suitable tech stack, build native iOS and Android apps that will have all necessary integrations and AI-powered feed.
- Help our client to perform a beta launch and continue development of a fully-fledged social media app.
Project team
- 2 UX/UI designers
- 2 Front-end engineers
- 2 React Native engineers
- 3 Back-end engineers
- DevOps
- 3 QA engineers
- Project manager
- Business analyst
- Software Architect
7 months for Beta Launch
4 months for Public Launch
Application features
Categories
Find and select news categories of interest
Profile
Manage info, content, reactions in one place
Video creation
Edit and upload original video and news content
AI-powered feed
Get relevant news selected by the algorithm
Community
Comment news and engage in discussions
AI supervision
Analyze content and detect violations fast
AI-powered feed
The idea of the app was to provide news from the sources a person trusts and generate relevant news recommendations. We implemented an AI-powered feed so that app users can not only get news recommendations but also view what’s trending.
A smart custom algorithm adjusts the news feed according to previous users interactions – news they read, videos they posted, reactions they left. So the users will not miss a trending thing worthy of their attention. The algorithm constantly learns and improves relevant suggestions.
AI content supervision
We’ve built a well-organized admin panel to manage all content with ease.
But we knew that with a huge number of posts, it is almost impossible to search for all violations and manage them manually.
To tackle this issue we decided to integrate cloud-based AI solution that can understand and analyze content. We stopped the choice on Hive, which helps to unlock insights into visual, audio, and text content with human-level accuracy and machine-level efficiency.
Hive looks for suspicious content and forms a list for removal that a human should approve. Besides, any user can report a post breaking the non-violation policy and it will be deleted.
Development approach
This app was designed with a large scale and complexity in mind. Therefore we selected microservices architecture.
- Most of the services were built on Node.js/Nest.js.
- For AI-related ones we used Python as a more suitable option.
- We used cloud services like AWS Lambda, AWS S3, AWS Elemental MediaConvert, etc. for specific tasks and also to improve scalability and performance of the app.
- For data storage and exchange we’ve used Amazon RDS and Amazon MQ.
Mobile application: Both iOS and Android apps were built from a single React Native codebase. Since React Native supports a huge library of functional and UI components, allows integration of native modules, and is compatible with most JS/TS libraries, it was a great choice for the app.
Web Application: Application has a public web UI for those who have no mobile app installed, and a web-based content management tool for the back office. We selected React for these web-apps with Next.js server-side rendering for SEO-friendly public content.
Outcome
Once the beta versions of the app were ready, our client successfully tested them on iOS and Android devices. We got a positive feedback, and our client received funding and invested it in further app development.
As of now, the app went through a public release. Although it’s too early to post an outcome, we’d like to share that the app shows some good numbers of user growth and engagement.
Our client has a lot of ideas and AI features in mind, and we will keep working together on making this app better and smarter.
