NVIDIA is releasing reusable agent skills and blueprints to help developers build and deploy vision AI systems that proc
NVIDIA announced tools for building vision AI agents that automatically turn video from factories, cities, warehouses and transportation systems into operational intelligence. The shift to edge AI is accelerating as enterprise workloads move closer to data sources. Gartner projects that more than two-thirds of enterprise-managed data will be created and processed outside the data center or cloud by 2028, and that over two-thirds of all enterprises globally will deploy edge AI by 2029, up from 10% in 2025. However, as much as 90% of existing edge data goes unprocessed. Building useful vision AI agents requires systems that can understand video, adapt to real-world conditions and connect insights to operational workflows while meeting latency, power, cost and connectivity requirements.
NVIDIA Metropolis agent skills and blueprints provide developers with reusable workflows to build, operate and optimize vision AI agents across their lifecycle. For simulation and synthetic data generation, Universal Scene Description, or OpenUSD, offers a common framework. Built on OpenUSD, NVIDIA Omniverse libraries help teams build simulation, synthetic data generation and digital twin workflows that model real-world environments across varying conditions such as lighting, weather, traffic patterns, camera angles, occlusion and rare events.
The announcement highlights three application areas. In manufacturing, Roboflow is integrating the NVIDIA Defect Image Generation skill and NVIDIA Cosmos world foundation models into its vision AI platform. For Corning's optical fiber manufacturing, a model trained on just eight real defect images augmented with synthetically generated data reached an average precision of 95% and perfect recall on the most challenging defect class, outperforming models trained only on real data and compressing a multi-quarter project into days.
In city operations, Linker Vision is building smart city systems with the NVIDIA Metropolis Blueprint for VSS to accelerate deployment of video reasoning agents across city infrastructure. In Kaohsiung, Linker Vision reduced development effort by 85% and reduced incident response times by up to 80%. Its newer AI-GRID expansion builds on this approach with NVIDIA NemoClaw blueprints for secure agentic AI.
In industrial environments, Foxconn and DeepHow use the NVIDIA Metropolis VSS blueprint with the Live Standard Operating Procedure Verification agent, where NVIDIA Cosmos provides the reasoning to interpret complex human activity. On NVIDIA GB300 server production lines, the solution improved first-pass yield by 3%, achieved 99% task-level accuracy in micro-action understanding and reduced redundant work by catching problems earlier.