In a major announcement at CES 2025, Nvidia unveiled its Blueprint for AI Agents that can analyse video, as revealed by CEO Jensen Huang during his opening keynote. The newly launched AI Blueprint, powered by the Nvidia Metropolis platform, aims to enhance productivity and safety across various sectors and even assists in improving sports performance.
Presently, over 1.5 billion enterprise-level cameras are operational worldwide, collectively generating an estimated 7 trillion hours of video each year. However, a mere fraction of this video data is effectively analysed, with less than 1% being monitored live by human operators. This significant underutilisation translates to substantial financial repercussions, particularly for the manufacturing industry, which incurs trillions of dollars in losses annually due to undetected product defects.
Interactive AI agents with advanced visual perception capabilities can act as continuous video analysts, improving operational efficiency, safety, and performance across various verticals. By deploying AI agents to process and analyse video data, organisations can recognise critical operational incidents that may otherwise go unnoticed.
To facilitate the development of these AI agents, Nvidia has launched early access to its updated Blueprint, which incorporates cutting-edge technologies such as Nvidia Cosmos Nemotron vision language models (VLMs) and Llama Nemotron large language models (LLMs). The framework allows developers to build AI agents that can efficiently process and analyse large volumes of video, achieving batch processing speeds that are reportedly 30 times faster than real-time viewing.
The Nvidia AI Blueprint also features several agentic capabilities such as task planning and tool calling, which then enhance the construction of versatile visual agents, applicable in solving various challenges. Furthermore, businesses can implement these AI agents in diverse environments, ranging from edge computing to cloud solutions.
The potential applications of video analyst AI agents are extensive. Within industrial operations, these agents can streamline productivity by ensuring adherence to standard operating procedures during complex processes. Additionally, they can optimise asset management by improving space utilisation in warehouses through 3D volume estimation, while also enhancing safety with auto-generated incident reports that ensure compliance with protective equipment guidelines.
In sports, a substantial market estimated at $500 billion, AI agents are expected to revolutionise how teams and coaches analyse performance. As illustrated by Huang during his keynote, an AI video analytics agent was capable of analysing an amateur baseball player's pitching technique, suggesting areas for improvement based on a comparison with professional skills.
The entertainment sector, valued at $3 trillion, also stands to gain from these advancements. Through Nvidia's Media2 initiative, AI agents will contribute to the generation of customised content, thereby aligning with individual viewer preferences and enhancing audience engagement.
Nvidia's Blueprint for AI Agents has sparked interest globally, with multiple partners, including Accenture, Infosys, and TATA Consultancy Services, beginning to integrate the blueprint into their development workflows. The initiative reflects a significant stride forward in harnessing AI's capabilities to transform business practices across various industries through enhanced video analysis.
Source: Noah Wire Services