TwelveLabs Launches Video-Understanding AI Agent 'Jockey,' Bolstering Full-Stack Strategy
TECHWORLD ·
✦ AI Summary
TwelveLabs has officially launched its AI agent, Jockey by TwelveLabs, based on its video-understanding foundation model. Jockey supports finding relevant scenes in video libraries through natural-language requests, along with explaining selection reasons, follow-up tasks, rough-cut generation, and social video analysis. TwelveLabs said Jockey connects the single workflow of video search, analysis, and editing and strengthens its full-stack strategy spanning from foundation models to agents.
TwelveLabs has officially launched its AI agent, Jockey by TwelveLabs, based on its video-understanding foundation model. With the launch, TwelveLabs aims to connect a single workflow for video search, analysis, and editing, while strengthening its full-stack strategy that spans from foundation models to agents.
Jockey can search for relevant scenes in a video library based on natural-language requests. It also explains why selected scenes were chosen and carries out follow-up tasks.
In addition to identifying needed scenes, Jockey can prioritize clips with strong response, such as those with high view counts and likes. It also supports generating rough cuts by linking desired segments from the original video.
The launch comes as the volume of video held by companies and creators continues to grow. In the process of searching for desired scenes and analyzing content, many cases have relied on ancillary information such as titles, tags, and subtitles, while much of clip selection and editing has still been done manually. As video volumes have expanded, the time required for search and organization has also increased.
TwelveLabs has streamlined that process with Jockey based on natural language. Users can invoke Jockey in the ChatGPT and Claude apps and carry out video tasks in a conversational format. The search context is designed to be maintained between ChatGPT and Claude. Developers can integrate Jockey into their own services and work systems through an API.
Jockey provides social video analysis features. Its input sources are YouTube, TikTok, and Instagram Reels links, and it analyzes related trends. For each video, it provides scripts, metadata, creator information, and engagement metrics, allowing users to compare search results that reflect both content and actual response.
Edited versions and analytical materials generated by Jockey can be stored and shared together with the video source. The output formats are edited MP4 files of the original video, social content analysis results, and video-based interactive websites. These outputs are used for team-level review and decision-making.
Before the official launch, TwelveLabs conducted a research preview for content producers that operate large-scale video archives. The official launch followed that research preview, and TwelveLabs incorporated user feedback collected during the process. It also added social trend analysis, performance metric integration, and output storage and sharing features.
The launch of Jockey marked an expansion in TwelveLabs' technical scope. The expanded scope ranges from video-understanding foundation models to an AI agent that carries out real work. TwelveLabs already had technologies for video search, embeddings, automatic metadata generation, summarization, and highlight extraction, and Jockey was built as a structure that connects those existing technologies. Based on natural-language requests, Jockey provides search, analysis, and editing results.
Lee Jae-seong, CEO of TwelveLabs, said Jockey is an AI agent that connects the company's core AI models. He added that Jockey extends video-understanding technology into real-world work. He also said the company plans to broaden its use cases so that vast video data can be connected to the information and outputs needed.
Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408336
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.