Jockey is a conversational video agent built on top of Twelve Labs API and LangGraph. It integrates the capabilities of existing Large Language Models (LLMs) with Twelve Labs' API, utilizing LangGraph for task allocation to distribute the workload of complex video workflows to appropriate base models. LLMs are used for logically planning execution steps and interacting with users, while video-related tasks are passed to the Twelve Labs API powered by Video Foundation Models (VFMs) for native video processing, eliminating the need for intermediary representations like pre-generated captions.