NVIDIA has announced the general availability of its AI Blueprint for Video Search and Summarisation (VSS), designed to empower organisations to build and deploy AI agents capable of processing large volumes of video data. This development aims to help businesses unlock insights from real-time and archived footage, addressing growing needs in sectors ranging from manufacturing and smart cities to sports and finance.
According to NVIDIA, video now makes up over half of all global data traffic. Yet, despite its prevalence, less than 1% of video content is currently analysed. With rising automation demands, workforce shortages, and the reshoring of manufacturing, video analytics is becoming an essential tool to bridge the physical and digital worlds.
The new VSS blueprint is powered by the NVIDIA Metropolis platform and integrates advanced AI capabilities such as vision language models (VLMs), large language models (LLMs), and retrieval-augmented generation (RAG). These technologies combine computer vision with language understanding, allowing AI agents to identify, summarise and interpret video footage with unprecedented speed and context.
AI agents delivering operational insights
The VSS blueprint enables businesses to summarise video up to 100 times faster than real-time viewing. For instance, an hour-long video can be condensed into text in under a minute. This capability opens the door for applications in worker safety, training, traffic management, and operational optimisation.
The system supports deployment on a single NVIDIA A100 or H100 GPU, and can also run on edge computing platforms like the NVIDIA RTX 6000 PRO and NVIDIA DGX Spark. It is capable of processing hundreds of live streams or short video clips at once. It also includes audio transcription, which is useful for interpreting meetings, training sessions, or other content where speech provides vital context.
By incorporating NVIDIA AI Enterprise software and NVIDIA NIM microservices, the blueprint provides a scalable and high-performance solution. With support for both visual and audio data, it offers comprehensive analysis across various enterprise use cases.
Real-world adoption across sectors
Several industry leaders have already deployed AI agents built using the VSS blueprint. Pegatron, a major electronics manufacturer, is using it to study best practices in assembly lines and reduce errors. Integrated into Pegatron’s PEGAAi platform, the Visual Analytics Agent monitors printed circuit board production and flags correct or incorrect actions. This implementation has helped reduce labour costs by 7% and manufacturing defects by 67%.
In Taiwan, the city of Kaohsiung is working with Linker Vision to improve emergency response across departments using AI-powered smart city applications. Previously siloed systems hindered coordination, but by combining real-time video analytics with generative AI, the new application can interpret complex events such as floods or traffic accidents. It currently supports 12 departments and will scale from 30,000 to over 50,000 city cameras by 2026, reducing response times by up to 80%.
The National Hockey League (NHL) has used the VAST InsightEngine with the VSS blueprint to search through petabytes of game footage. This enables near-instant retrieval of highlights and automates content creation by tagging and assembling clips. Future possibilities include generating dynamic insights on player performance and match strategies during live broadcasts.
Siemens is leveraging the VSS blueprint in its Industrial Copilot for Operations. This AI assistant helps factory workers with equipment maintenance, troubleshooting, and performance enhancement. Built with VLMs, LLMs, and NeMo microservices, the copilot has increased productivity by 30% and has the potential to push it up to 50%.
Growing ecosystem of partners
NVIDIA’s blueprint is also being embraced by a wide ecosystem of partners across sectors. Superb AI rolled out a system at Incheon Airport to cut passenger wait times within weeks. In Malaysia, ITMAX is applying the technology to manage city infrastructure in Kuala Lumpur more efficiently.
In advertising, PYLER used the VSS blueprint to integrate brand safety and targeting tools, helping clients like Samsung and BYD achieve stronger ad performance. BYD, for example, reported a fourfold increase in click-through rates. Financial brand Hana also exceeded its campaign targets by using more relevant and positive content alignment.
Meanwhile, Fingermark is embedding the VSS blueprint into its Eyecue platform to serve quick service restaurants. The enhanced platform will provide actionable data on drive-thru wait times, staff performance, and service bottlenecks.
With widespread use cases already proving its value, the VSS blueprint looks set to accelerate the creation of video analytics AI agents across sectors, enabling faster decision-making, safer operations, and more efficient processes.
