Snowflake Ventures Invests in Twelve Labs to Bring Advanced Video Understanding to the Snowflake AI Data Cloud for Media

In a rapidly changing and competitive media and advertising industry, media companies, sports organizations, advertising agencies and others are consistently looking for ways to improve the consumer experience and drive monetization. This includes content analysis, video and creative search capabilities, content personalization and creative versioning. These capabilities are all unlocked by leveraging the joint AI capabilities now available from Snowflake and Twelve Labs’ video embeddings. These embeddings also help machine learning algorithms process complex data and the links between data objects, enabling the core features that Twelve Labs is bringing to market.

Twelve Labs provides state-of-the-art multimodal embedding technology

Twelve Labs’ Video Understanding Platform uses AI to extract visual, audio, textual and contextual data from videos, enabling semantic search, analytics and insights at scale. Its foundational models are among the leaders in video description creation, scene detection and logical breakpoint designation — allowing videos to adapt to different device screen sizes while maintaining video quality. Traditional approaches rely on frame-by-frame analysis or separate models for different modalities. Twelve Labs’ API captures the complex interactions between visual cues, body language, speech and overall video context by generating contextual vector representations of these interactions. This holistic approach enables media organizations’ data teams to work with a more nuanced understanding of video content.

Earlier this year, Snowflake and Twelve Labs partnered to make the Twelve Labs Video Understanding API available to all AI Data Cloud users. With Twelve Labs and Snowflake, customers can bring the AI and video embeddings workflow to their data. This helps them maintain strong privacy and governance over their video data and embeddings while benefiting from advanced video analysis capabilities, all within a secure environment to analyze visitor behavior and reactions. Users can also locate specific footage from huge content repositories and reuse that data to generate new content that can be targeted at consumers based on data in Snowflake.

The Twelve Labs API brings humanlike video understanding to any application, even with terabytes or petabytes of video. These rich, multimodal video embeddings can then be stored in Snowflake using its vector data type support. This enables more sophisticated video analytics and AI-driven video-based applications within the AI Data Cloud. This integration leverages Snowflake Cortex AI, Snowflake’s fully managed AI service that provides a suite of generative AI features, and Snowpark Container Services for scalable processing. Developers can create custom applications combining Twelve Labs' video understanding technology with Snowflake's data management capabilities, enabling advanced video analytics, search, content moderation and recommendation systems. This powerful combination allows organizations to build sophisticated, AI-driven solutions for video-centric applications while benefiting from Snowflake's robust data platform.

Following our investment, Snowflake and Twelve Labs will work together to identify opportunities to make it even easier to bring powerful video AI capabilities into a unified data platform.