Lumana Integrates NVIDIA Metropolis Blueprint for Video Search and Summarization (VSS)

The Problem: Cameras Everywhere. Answers Nowhere.

The world is blanketed in video. Enterprises, cities, retailers, and industrial operators run thousands of camera feeds simultaneously — and yet, the moment something happens, the question is always the same: what actually occurred, and do I need to act?

Traditional video security was built to store and record, not to perceive, reason, and respond. It created a network of cameras and an infrastructure without a critical intelligence layer. The result is a compounding effect of false alarms, missed events, and operator fatigue. Key issues include:

• Alert Overload: security teams are inundated with notifications — the vast majority are false positives. When everything is an alarm, nothing is.
• Zero Context: legacy AI can tell you something moved. It cannot tell you whether a retail customer bypassed checkout, whether a worker is in danger, or whether a facility is truly clean.
• Slow to Act: when operators must manually review clips to validate an event, critical response windows close. Speed matters — and manual review doesn’t scale.
• Unexplainable Outputs: existing systems give you a flag and a timestamp. They don’t explain why something was flagged or what the recommended action is — leaving operators to guess.

The Solution: Traditional Video Security vs What Lumana + NVIDIA VSS Unlocks

Traditional video security was built to record and react — not to reason. It generates alerts without answers, flags events without context, and leaves operators to manually bridge the gap between what a camera sees and what it actually means.

Lumana and NVIDIA VSS close that gap. By applying AI across the entire video security experience — from real-time detection to contextual scene understanding to verified, decision-ready output — organizations move from drowning in alerts to acting on validated insights. The result is sharper detection, dramatically fewer false positives, and a system that scales without adding headcount or hardware.

How it Works: Multi-Stage Intelligence Pipeline Built for the Physical World

Lumana’s platform combines continuous-learning AI, edge-based detection, and centralized reasoning into a single platform — creating a closed loop between detected and validated, decision-ready insights.

“Object detection creates noise. Activity and behavior detection closes the gap — but what operators truly need is more context and understanding. We’ve built a system that doesn’t just alert teams to events. It tells them what happened, why it matters, and what to do next. That’s the intelligence gap we’ve closed.”

— Sagi Ben Moshe, Founder & CEO, Lumana

From retail floors to factory lines to restaurant kitchens, Lumana + NVIDIA VSS delivers verified insight where it matters most.

Intelligence That Works Across Every Industry Vertical

From retail floors to factory lines to restaurant kitchens, Lumana + NVIDIA VSS delivers verified insight where it matters most.

Delivering Real World Impact with Lumana and NVIDIA VSS

Lumana’s partnership with NVIDIA doesn’t stop at commercial or enterprise environments. Through its centralized, GPU-accelerated video sensor system, Lumana has converted thousands of traffic cameras into real-time intelligence nodes — without deploying a single piece of edge hardware at any intersection.

Live video streams are processed through a centralized NVIDIA-accelerated pipeline, producing structured telemetry — vehicle counts, congestion indices, speed metrics, safety indicators, and environmental analytics — all delivered into the Lumana Insight Data Portal for transportation agencies and municipalities nationwide.

Beyond traffic, this infrastructure generates predictive air quality indicators by correlating vehicle class distribution, idle duration, and density fluctuations across corridors — transforming existing roadway infrastructure into a dynamic environmental sensing network with no additional physical sensors required.

Why NVIDIA is The Right Technology Partner for What We’re Building

Lumana selected NVIDIA VSS not because it’s a market leader — but because it solves a specific, hard problem: the gap between detecting something and truly understanding it.

Cosmos-Powered Reasoning

NVIDIA VSS uses Cosmos world foundation models that go beyond object classification. They reason over scene context, temporal sequences, and environmental factors — the same dimensions that humans use to evaluate whether a situation is genuinely risky or a false alarm.

Production-Grade GPU Acceleration

Lumana processes thousands of concurrent video streams simultaneously. NVIDIA AIinfrastructure enables high-throughput, low-latency inference at scale — maintaining consistent accuracy across jurisdictions without degradation.

Structured, Explainable Outputs

VSS doesn’t return a probability score. It returns a structured package — event summary, risk classification, and reasoning chain — that operators can act on immediately, and that auditors can review with confidence.

Software-Defined Deployment

NVIDIA’s centralized compute model aligns perfectly with Lumana’s architecture: no hardware required at every camera or intersection. Activate AI across existing infrastructure instantly — from a single building to a national camera network.

Available Now. Showcased at NVIDIA GTC 2026.

Lumana’s AI-powered video intelligence platform, integrated with NVIDIA VSS, is available today for enterprise, government, and commercial deployments. Lumana will be demonstrating its full VSS-powered capabilities at NVIDIA GTC 2026.

Whether you operate a network of retail stores, an industrial facility, a restaurant chain, or a municipal infrastructure portfolio, Lumana and NVIDIA VSS deliver the same result: cameras that don’t just record video but understand what they’re seeing.

Ready to Transform Your Video Security into Video Intelligence?

See what Lumana and NVIDIA VSS can do for your organization. Request a demo and discover how much value can be unlocked from your video data