CINDER · 2026
Case study
How Zello Catches 3x More Repeat Offenders with Cinder
A push-to-talk app with millions of users replaced manual moderation sweeps with daily automated detection across languages, harm categories, and account-creation events.
6 MIN READ
The stakes
When trust and safety becomes top of mind
Zello is a push-to-talk voice communication platform serving millions of users worldwide. The product spans two surfaces: Zello Work, the paid product used by frontline workers in retail, aviation, hospitality, and other industries, where customers manage their own private networks, and Zello Friends & Family, the free, social, user-generated-content side of the platform.
Following heightened public scrutiny of how bad actors were using the open, public network to organize, the company committed to taking action against them on its public Friends & Family network, formalizing trust and safety as a dedicated function and investing in a program built to detect and act on emerging threats systematically. As with any open, global user-generated-content platform, the Friends & Family network includes a wide range of user behavior, requiring ongoing detection and enforcement.
The challenge
To run a program that could operate at the speed of the threats it was trying to catch, Zello needed infrastructure built for the job.
The first version of the Zello Friends & Family Trust and Safety program ran on Zello Tools, the company's existing internal moderation system. The legacy tool covered the basics: a list of reported users, an image review queue, and a search across channel names and descriptions. However, there was no way to view content during an appeals review, no real decision audit trail, and no way to do proactive detection without leaving the tool entirely.
Proactive work happened manually outside the tool, with abuse-term watchlists tracked in Google Docs, and sweeps for ban-evading accounts and channels named after extremist or criminal acronyms running once or twice a week. Isabella Waltz, Zello’s Head of Trust and Safety said this could take up to eight hours in a single sweep. “And that was just treading water,” she says. “That was not ever expanding the scope.”
Zello’s problem wasn’t its strategy but its speed. Repeat offenders on a free, international communications platform routinely cycle through accounts faster than a manual review can absorb. To run a program that could operate at the speed of the threats it was trying to catch, Zello needed infrastructure built for the job.
It's not just a lot of time saved. It's the ability to reinvest that time into higher-impact moderation work and expand coverage where it matters.
The approach
From Manual Sweeps to Daily Detection
Zello's Cinder integration immediately delivered results the legacy system never could. The platform’s first true moderation infrastructure went live within weeks, Zello's team could finally review appeals against the actual content in question, and Cinder captured every decision with a complete audit trail, the foundation Zello is now using to be more efficient in addressing compliance obligations.
The operational shift came when Zello's engineering team wired account-creation and channel-creation events into Cinder. With those signals flowing, the manual sweep workflows that had previously consumed entire workdays collapsed into automated Cinder workflows running every day.
Repeat-offender detection at account creation.
The manual ban-evasion sweep that used to consume up to four hours of moderator time per week now runs automatically inside Cinder. In its first 30 days of operation, the automated workflow produced 626 enforcement decisions with no manual sweeps required, a roughly threefold increase in repeat-offender takedown volume.
Proactive channel surveillance.
Pre-Cinder, abuse-term sweeps across channel names ran weekly through a buggy legacy search. In Cinder, those terms now live as managed keyword lists feeding bulk review. Of roughly 500 channels flagged in a typical week, about 30 require human review. Zello's moderation team now runs this workflow in about an hour, almost daily, instead of the once-or-twice-weekly scramble it used to be.
The deeper unlock? Coverage.
With the manual cost of running new categories collapsed, Zello has expanded what it watches for: abuse-term coverage in different languages, detection signals for organized harmful activity, and emerging harm categories added within days of identifying them. Each addition compounds against the same set of automated workflows, so every new term Zello adds reaches its entire Friends & Family user base immediately, in every language, every day.
The pay off
Results at a Glance
3X
More repeat-offender accounts taken down vs. the legacy manual-sweep workflow.
2X
More abusive channels detected, growing further as new harm categories are added.
50%
Of all bans now executed by orchestrated workflows triaging urgency and threat level.
What's next
Voice, compliance, and agents.
Zello and Cinder are now mapping a six-to-twelve-month roadmap aimed at three priorities, each one targeting a coverage gap that, until the Cinder integration, was operationally out of reach.
Voice moderation: Zello is, fundamentally, a voice product, and moderating voice content at scale on the Friends & Family network has historically been out of reach for a small team. Building on Cinder's infrastructure puts that capability on the horizon for the first time.
Compliance, end-to-end: Zello is producing its DSA transparency report for 2025, working from the audit trail Cinder has been generating since signature. The exercise has turned into a map of every remaining gap in the program, and most of those gaps map directly to capabilities within Cinder. The next year of compliance work is now an activation question, not a tooling one.
Agentic moderation: Following the proven pattern Cinder customers like Character.ai have deployed, Zello plans to introduce AI agents on high-confidence, low-ambiguity classification tasks. The goal is the same one that has shaped every phase of the Cinder rollout: free the moderation team from routine work so it can focus on the strategic, sensitive, and nuanced cases where human judgment is irreplaceable.
Bring this rigor to your network
Talk with the team that helped Zello triple its repeat-offender takedowns and automate half of all bans on a free, global voice network.