Curio powers your
Because You Watched
It matches viewers to how a title feels to watch, not just its genre, and puts the right next title where it counts: after the credits, on the home screen and on every title page.
Weak discovery has a real cost.
Average time viewers spend searching for something to watch
Abandon the session when that search comes up empty
Say they would cancel a service over trouble finding something to watch
Source: Gracenote, 2025 State of Play (November 2025). Survey of 3,000 consumers in Brazil, France, Germany, Mexico, the UK and the US.
That’s rarely an algorithm problem. It’s a foundation problem. Most recommendation systems learn from sparse inputs: genre, cast, release year and a handful of tags. So two thrillers look identical, even when one is a slow-burn character study and the other is nonstop action.
Less searching. More watching.
Recommendations that feel like them
Matched on pacing, tone and emotional arc, not just the genre a title is filed under.
Less scrolling to find something
The rows already make sense, so viewers spend their time watching, not browsing.
Titles they would never have found
New releases and long-tail titles get recommended from day one, before anyone has watched them.
One foundation. Three places viewers feel it.
Keep the session going when the credits roll
When a title ends, Curio lines up what to watch next based on what other viewers who loved that same title went on to enjoy. It’s the natural next pick, not just a look-alike.
Rows that fit each viewer from day one
Even a brand-new viewer who has only watched one or two titles gets home-screen rows that fit, shaped by pacing, tone and story rather than only the genres they’ve clicked.
Smarter title detail pages
On every title page, Curio suggests titles that feel like the one in front of the viewer, connected by tone, pacing and emotional arc rather than category alone.
Matched on how a title feels, not which box it’s in.
Curio runs natively on Vionlabs multimodal embeddings, not metadata. They capture each title’s visual composition, pacing, dialogue, tone and emotional arc, learned from more than 1,000,000 analyzed assets. A two-tower model, the approach behind personalized feeds at TikTok, YouTube, Meta and Instagram, matches viewer and content in one shared space.
Privacy-first, and quick to switch on.
Curio works on pseudonymized viewer IDs only. No personal identifiers need to leave your environment, and the integration is built to keep you fully GDPR-compliant.
Watch History API
Fast, stateless integration.
Full deployment
Long-term behavioral profiles for deeper personalization, plus expanded recommendation types and performance analytics.
Live today with Great American Pure Flix.
Vionlabs has powered Pure Flix’s editorial curation for several years. Pure Flix is now the first platform to bring Curio to its members, across all three experiences: Personalized Recommendations, Because You Watched and Similar Titles on title detail pages.
Read the press release →“This isn’t ‘recommendation’ as usual. It’s the next expression of what our multimodal embeddings already understand about content. We built the deepest content understanding layer in the industry, and personalization was always the natural next step. Now that same understanding carries straight through to what a viewer actually sees on their screen.”
