Twelve Labs said on Thursday it has officially launched 'Jockey', an AI agent that performs tasks from video search to analysis and editing.
Jockey runs on a video-understanding foundation model developed in-house by Twelve Labs. When users make requests in natural language, it finds the needed scenes in a video and explains the basis for the selection. It also selects key clips and produces rough edits.
If users enter links to social videos on YouTube, TikTok or Instagram Reels, it can also analyze trends and engagement indicators such as views and likes. The analysis results and edited versions can be saved and shared along with the source of the video.
Users can call up Jockey in the ChatGPT or Claude app and perform video tasks as if chatting. Developers can connect it to their own services or work systems through an API.
Lee Jae-sung (이재성), chief executive of Twelve Labs, said, "Jockey is a next-generation AI agent that organically connects Twelve Labs' core AI models and takes video understanding technology to a new level," adding, "Anyone will be able to draw practical value from vast video data."