Using AI to reduce fraud and spam 175 million monthly listeners
Think-it developed data-driven AI models to identify fraudulent activity, spam, and bot engagement.
The story
As a musical platform, the trust and the safety of both music creators and listeners is vital.
Users should feel safe when using the platform.
- Creators should receive authentic engagement.
- Internal teams should be able to take informed actions towards suspicious activity.
- Payment teams should only pay for authentic plays.
Think-it’s role
We worked across different data owners to start consuming and producing new data to inform and train the AI model for fraud and spam. We were able to** reduce spammy comments 70-80%**, which allowed the team to redirect focus to other areas. We also reduced the need for ad-hoc analysis requests from daily to every two weeks. This enabled better human decision making as well as reinforcing existing and new features.
Why it mattered
They were able to solve the issue of spammy comments on the platform while also significantly increasing the human efficiency of the Trust & Safety team. This is keeping both creators and listeners satisfied, while ensuring the efficiency of the internal teams.
Testimonials
What our partners say about us
Work with a similar team
Related Articles/Press
Atomic Design System in Jetpack Compose
266 days agoOur guide on creating an atomic design system using Jetpack Compose
Atomic Design System in SwiftUI
266 days agoWe will explore the fundamentals of SwiftUI and how to leverage atomic design methodology to create scalable and maintainable design systems.
UI Performance Rendering
265 days agoThe rendering process relies on multiple system parts, and each task might lead to rendering issues that can point to the necessary optimizations.
