TwelveLabs Launches Jockey, an AI Video Agent That Lets ChatGPT Search Inside Footage

Video intelligence startup TwelveLabs officially launched its new service “Jockey” on October 15, positioning it as an orchestration layer that sits on top of the company’s Marengo and Pegasus video‑understanding models. After a research preview with large media customers, Jockey is now being opened up more broadly as a way to search and reason over massive video libraries.

Organizations upload their footage to the cloud and can then ask natural‑language questions like “show me every scene where this spokesperson appears” or “find clips where the product malfunctions on stage.” Jockey parses the request, locates the right segments, and explains why they match. A key twist is that Jockey is accessible directly from popular chat interfaces such as ChatGPT and Claude, so users can stay inside their existing AI apps and still run deep video queries or request summaries of specific series or episodes.

With its public debut, TwelveLabs is expanding developer APIs and cross‑app workflows, aiming well beyond traditional broadcasters. The company is pitching Jockey as infrastructure for any workflow where video has been a black box: corporate training libraries, security footage review, and creator asset management. In essence, it’s an attempt to make hours of footage as searchable and interactive as text documents.