Handling fast-moving interactions: Custom moderation for Wizz’s Gen Z-first app

Wizz App is a social discovery platform for Gen Z users seeking real connections through chat, casual games, and friend-finding features.

As the platform rapidly scaled to 16M+ users, maintaining a safe environment became critical. In a context where communication trends evolve quickly, the challenge was to moderate high volumes of real-time conversations while preserving fluid and authentic interactions.

At the end of 2024, Wizz App partnered with Bodyguard to strengthen its moderation capabilities and better handle a growing complexity: distinguishing genuine threats from slang, sarcasm, and the shorthand Gen Z actually uses.

A Gen Z-first platform

A Gen Z-first platform, Wizz App's core audience is made up of 17–24 year olds.

To manage safety across age groups, the platform enforces strict age segmentation, ensuring users in younger age brackets and adults interact in separate environments with different levels of risk and moderation requirements.

The younger audience relies on informal language, exaggeration, and fast-moving trends, making intent harder to interpret, while being more vulnerable to grooming, exploitation, and peer-based harassment

Operating at scale

• 3M+ monthly active users

• 100 % Gen Z

• 810M+ messages exchanged each month

Complex conversations at scale

Wizz App connects young people in real time, where harmful content can surface and spread quickly. Moderating at this scale means handling a high volume and variety of messages across different age groups, each requiring context-aware review.

At the same time, communication patterns evolve rapidly. Gen Z relies on informal language (“u”, “idk”), exaggeration (“I’m dead”), and constantly shifting trends, making intent harder to interpret and increasing ambiguity in conversations.

In this context, traditional moderation approaches based on keywords struggle to keep up since these generic moderation tools were not built for the nuances of how Gen Z communicates. A filter that flags "I'm dead" as a self-harm risk, or reads "u want beef?" as a food-related, misses the point entirely. Generic approaches fail to distinguish between age-appropriate banter and genuine harassment, and do not reliably detect more complex risks such as off-platform solicitation or the sharing of sensitive information.

Context-aware moderation

Wizz App was looking for a moderation partner capable of building models tailored to its platform, rather than relying on generic tools. This meant adapting to its language, user base, and risk profile, while operating at scale and continuously refining classifications based on real moderation feedback.

Working with Wizz App, Bodyguard adapted its existing and already extensive classification framework to the platform’s specific needs, extending it to better cover the wide range of behaviors observed, from harassment and hate speech to self-harm, scams, or the sharing of sensitive information. The system is both age-adaptive and context-aware: it evaluates the severity of each message, identifies the intended recipient, and adjusts decisions based on the context of interactions. Critically, all of this happens before a message is delivered, and harmful content is intercepted in real time, so users never see it.

As Alexandra Ryabova, Head of Operations at Wizz App explains, “Most moderation tools are built for adult social networks and apply broad category filters. Bodyguard worked with us to build something calibrated to Wizz App, adapted to our user demographics and the specific risks that come with connecting younger users.” 

This approach is reinforced by a continuous feedback loop, where real-time moderation inputs from Wizz App teams help refine the models over time, improving performance as new behaviors and risks emerge.

Results

• 51% reduction in toxic messages over one year

• +47% increase in recurring users

• 86% of users report feeling safe meeting people online

"We've tested a lot of tools. Bodyguard built this with us, around our users, our language, and our risk profile. The classification depth reflects that. And when something doesn't work, they're as dedicated to tackling it as we are. That level of investment is hard to find" said Alexandra Ryabova, Head of Operations at Wizz App.

Conclusion

With Bodyguard, Wizz App is able to moderate high volumes of real-time interactions across diverse age groups, using context-aware models to better detect complex risks such as off-platform spam while maintaining a fluid user experience.

Fully aligned with Wizz Apps' safety-first approach, the solution integrates into a broader strategy where trust and safety are central, representing a significant share of the platform’s operational focus and investment.