Understand your catalogue
Deep understanding of the content itself
Rank on what is actually in an item - scenes, actions, dialogue, sound, objects and themes - rather than on its title and tags.
What it does
The mechanism
- 01
Multimodal embeddings let ranking read the content itself. TwelveLabs' Marengo model covers video, audio, image, text and document embeddings; the same pipeline takes text embeddings from any provider.
- 02
Because relevance can be judged from the content, a new item can be placed the moment it is published, before it has accumulated any interaction history.
What it gets you
Three things you can point at
01
Beyond titles and tags
Two items with unrelated metadata but the same subject, tone or setting are recognised as related.
02
Cold start, solved properly
New and long-tail items are ranked on what they are, not on how many people have already clicked them.
03
Find invisible content
Good items that rarely get an impression, because the current system cannot confidently place them, are surfaced rather than left to sink.