Twelve Labs

Twelve Labs is a video AI company founded in 2021 by Aiden Lee, Jae Lee, Dave Chung and Sungjun Kim, with headquarters in San Francisco and additional offices in Seoul, New York and London. The company builds multimodal foundation models that let computers understand video the way humans do, by combining visual, audio and language information rather than relying only on manual tags.

Its product line centers on two models: Marengo, which creates embeddings for natural-language video search across more than 47 languages, and Pegasus, a video-language model that reasons over footage up to two hours long to generate summaries and insights. Organizations in media, advertising and the public sector use the platform to search, segment and analyze large video libraries without manual metadata tagging.

1. Natural-language search: Finds specific moments in video libraries using plain-text queries.

2. Marengo embeddings: Generates multimodal embeddings across visual, audio and language data.

3. Pegasus reasoning: Analyzes footage up to two hours long to produce summaries and insights.

4. Content segmentation: Automatically splits long-form video into structured, searchable segments.

5. Compliance screening: Flags content for brand-safety and moderation review.

6. High-throughput processing: Processes video at roughly 60 times real-time speed at scale.