Details
- NVIDIA AI announces the Metropolis Video Search and Summarization (VSS) 3 Blueprint, focused on video search and summarization agents.
- The release adds 16 new skills for coding agents, including search, summarization, alerting, and report generation.
- Coding agents can now analyze massive live video streams and large video libraries using simple natural language prompts, reducing manual configuration.
- VSS 3 builds on NVIDIA Metropolis and NVIDIA NIM microservices, combining vision-language models, large language models, and retrieval components for multimodal video understanding.
- The blueprint’s modular architecture and reference workflows support diverse environments, from smart cities to warehouses, and can summarize long-form video up to 100x faster than manual review.
- Agentic search capabilities let AI agents decompose complex queries and search across video embeddings to locate specific events, actions, and object attributes in seconds.
- Early access updates introduce real-time alerts, alert verification, long video summarization, and automated report workflows, plus expanded hardware support across edge and data center GPUs.
- Developers can use coding agents augmented with VSS skills to deploy profiles, index video, run search and summarization, configure alerts, and integrate outputs into custom applications through an agentic chat interface.
Impact
By packaging video understanding, search, and summarization into a modular, agent-ready blueprint, NVIDIA lowers the barrier for enterprises to deploy video analytics AI at scale. This strengthens Metropolis as a platform for smart infrastructure and narrows the gap with broader AI agent ecosystems by making multimodal, edge-to-cloud video intelligence easier to automate and integrate.
