Building trust: the story behind Streamr's family-safe pipeline
Three weeks before launch, a church pastor in Ohio emailed us. 'We need to go live on Sunday,' she wrote. 'But we need to know our kids won't see garbage in the chat.' That email sat on my desk for two days. It was the moment I realised we'd been building a feature; we needed to build a system.
What family-safe actually means
When we started Streamr, 'family-safe' was just a word we threw around. It sounded good. But the Ohio pastor's question forced us to define it properly. Family-safe doesn't mean sanitised. It doesn't mean removing all disagreement or controversy. It means creating a space where parents don't have to police every second of their kid's screen time, where a 12-year-old can watch a worship stream without encountering the worst of the internet's underbelly.
That's a specific problem. It needs a specific answer. We spent weeks talking to church leaders, parents, and content creators. The pattern that emerged was clear: they wanted moderation that was consistent, transparent, and not hiding behind an algorithm nobody could explain. They wanted rules they understood, applied fairly, across the entire platform.
That requirement shaped everything that came after.
The temptation to outsource
The obvious move was to bolt on a third-party moderation service. There are plenty out there. They're designed for exactly this. We looked at five different options in the first month.
But every demo felt the same. Good at scale. Generic. Built for platforms where the creator doesn't know the moderator, and the moderator doesn't know the community. Built for TikTok, basically. That's efficient. It's not what Streamr needed.
A pastor streaming a Sunday service to 200 people they've known for years is not the same as a viral creator reaching 200,000 strangers. One comment in the live chat at a church stream isn't a data point; it's a voice in your community. Applying industrial-scale moderation to that felt wrong.
We decided to build something ourselves, but keep it human-focused. Not a filter you toggle on and forget about. A pipeline we could calibrate, see through, and defend to our creators.
The pipeline we built
Here's what we settled on. Three layers, each with a different job.
First: automated detection. We catch the obvious stuff. Slurs, spam patterns, explicit material. Not because we think machines are perfect at this, but because they're fast and consistent at finding clear violations. This layer is transparent. Creators and churches can see what it caught and why.
Second: community flagging. Viewers can report content directly. This gives the community a voice. It also gives us signal we can't get from automation alone. A comment might pass every filter and still make someone uncomfortable. We needed to hear that.
Third: manual review by real people. Not an algorithm deciding in secret. When something lands in the grey zone (sarcasm, context, tone), a person looks at it. They have access to the stream context, the creator's history, the community involved. They make a call based on what Streamr is, not what generic social platforms think they should do.
Streamr Kids, our curated child-safe content area in Streamr Plus, sits on top of all three. Content has to pass the moderation pipeline and then be explicitly marked as appropriate for younger viewers. It's an opt-in layer, not a filter you fight against.
The decision we had to make
About six weeks in, we faced a real test. A stream included a discussion of a tough Bible passage. The content wasn't explicit. But it was complex. Several viewers flagged it. One said their 10-year-old asked a question they weren't ready to answer yet.
We didn't remove the stream. The creator was right to discuss it. But we moved it out of Streamr Kids. The reasoning was straightforward: not because it was wrong, but because context matters. That same stream made perfect sense for a youth group or a family that'd been preparing the conversation.
That decision told us we were on the right track. We weren't trying to impose one family's values on all families. We were creating options. Creators could stream what they needed to stream. Families could choose what made sense for them. The pipeline was there to prevent actual harm, not to control content.
What we learned
The hardest part wasn't the technical work. It was accepting that perfect moderation is impossible. Some content will get through. Some good content will be flagged. The goal isn't perfection. It's accountability. When something happens, we know who to talk to, why it happened, and what comes next.
We also learned that creators want to be part of this. Churches especially. They don't want moderation done to them. They want to help shape it. We built tools for that. Creators can set their own chat guidelines. Churches can configure what 'family-safe' means for their community. The moderation pipeline is strong, but it's not rigid.
That's the difference between a filter and a system. A filter says 'yes' or 'no'. A system says 'here's how we think about this, here's what we've seen, here's what you can do about it.'
If you're a creator or a church that's been locked out of other platforms, or you've just watched your community get worse by the month, the question isn't whether you want moderation. It's whether you want a say in how it works. Does that matter to you?
Ready to try Streamr by MRVL?
One tap to download. No sign-up wall.