12Labs builds foundation models for video, enabling human-like retrieval and generation of text output from video content. Traditional tagging methods are limited and inflexible, while 12Labs' approach allows for semantic search and retrieval of video assets.