SANTA CLARA — NVIDIA has unveiled a new version of its AI Blueprint for video search and summarization, powered by the Metropolis platform, offering a transformative approach to AI agent development. This powerful tool allows organizations to unlock new levels of productivity and safety by enabling real-time video analysis, transforming the vast volumes of video footage generated by enterprise cameras into actionable insights.

In 2024, over 1.5 billion enterprise cameras worldwide are capturing about 7 trillion hours of video annually, but less than 1% of this content is reviewed by humans. This underutilization presents a massive opportunity for AI, as businesses lose trillions due to undetected operational issues or product defects. By integrating AI agents with visual perception, NVIDIA’s new blueprint helps fill this gap, allowing industries to automate video analysis, improve efficiency, and increase safety.

The blueprint, which incorporates NVIDIA’s Metropolis platform, Cosmos Nemotron vision language models (VLMs), Llama Nemotron large language models (LLMs), and NeMo Retriever, significantly accelerates the process of analyzing vast quantities of video and images. Developers can leverage the toolkit to create AI agents capable of batch video processing up to 30 times faster than real-time viewing. This suite of models includes features like chain-of-thought reasoning and task planning, which streamline the creation of robust AI agents for a range of applications.

NVIDIA’s AI agents are poised to revolutionize industrial operations by optimizing workflows, enhancing productivity, and improving safety. These agents can track complex processes, ensure compliance, and prevent accidents by identifying anomalies in real-time. They can also assist with asset management, improving space utilization and optimizing inventory in warehouses through advanced visual capabilities like 3D volume estimation.

The technology extends beyond industrial applications. In the sports and entertainment sectors, AI agents are set to make a significant impact. With industries such as sports analytics and media benefiting from real-time video analysis, teams can leverage these tools to enhance player performance and boost fan engagement. During his CES 2025 keynote, NVIDIA CEO Jensen Huang showcased how an AI video analytics agent analyzed his own fastball pitch, offering insights into areas for improvement based on video analysis.

Additionally, NVIDIA’s Media2 initiative aims to improve content creation in the media industry by providing more personalized and impactful experiences for audiences. With widespread adoption across global partners, including Accenture, Deloitte, Infosys, and others, the AI Blueprint for video analysis is already being integrated into various developer workflows to drive innovation across industries.

Apply for early access to the NVIDIA Blueprint for video search and summarization.

See notice regarding software product information.