Jockey is an AI-powered video intelligence agent from TwelveLabs that lets users search, organize, and extract insights from entire video libraries using natural language. Built on TwelveLabs' Marengo and Pegasus models, it can identify people, objects, moments, and sentiment within footage without relying on filenames or manual tags. Marketing teams can use it for ad tagging and hook detection, while developers can access multimodal embeddings and structured data extraction via API. Jockey integrates with Claude through MCP and offers frame-accurate timestamps for every search result.