Lumana AI Inc., a startup using artificial intelligence to automate video surveillance, wants to alter the way businesses monitor the physical world after raising $40 million in an early-stage funding.

Today’s Series A round was led by Wing Venture Capital and saw continued backing from Norwest Venture Partners LP and S Capital VC Management Ltd., which led its initial, seed funding round back in April 2024. To date, the company has raised $64 million in funding.

The startup is building an AI-native video surveillance platform that transforms security cameras into intelligent and context-aware recognition systems. It’s capable of identifying threats and understanding human behavior, and can even initiate automated responses to incidents it spots in real time. It’s designed to spot incidents such as violence, theft, fire, unauthorized access, safety violations and more, and can even generate alerts before any of those incidents occur.

The secret behind these capabilities is Lumana’s specialized vision-language models or VLM, which are paired with agentic AI capabilities to enable video feeds to become proactive stakeholders in safety and security operations.

As Lumana founder and Chief Executive Sagi Ben Moshe puts it: “We’re building the infrastructure for cameras to perceive, understand context and take action, not just watch.”

Lumana stands out particularly for its integration of AI agents, which are specialized AI systems that can automate various security and safety-related tasks and workflows without human input. In Lumana’s case, its intelligent AI agents are geared towards specific roles such as identification, tracking, compliance monitoring, safety enforcement and so on, making operational decisions without human supervision. They can also be integrated with third-party safety and security systems, the company said.

The startup said its AI agents can generate alerts for multiple threat scenarios and accidents that require a response, such as if they identify someone with a weapon, or if they spot a fire or see someone fall and injure themselves, or if an intruder gains unauthorized access to a restricted area.

The system is overseen through a customizable dashboard that transforms video surveillance into actionable insights, and it can also be used for investigating incidents, with more than 100 searchable attributes.

Lumana, which was founded in 2021 and emerged from stealth last year, says its platform is already being used by McDonald’s Corp., Meta Platforms Inc., the Minnesota Twins basketball team and New York University. It reckons its cloud-based platform processes more than a billion images per day on behalf of its customers, while helping human security and safety teams to review thousands of hours of video footage in seconds.

A more recent use case for Lumana’s platform is operational optimization. For instance, the laundromat chain SpinXpress Laundry uses it to analyze customer dwell times so it can optimize staffing, to ensure its customers waste as little time as possible.

Wing Venture Capital Founding Partner Peter Wagner said Lumana is about providing real-time intelligence on physical world conditions. “The system does what traditional surveillance systems cannot,” he said. “It learns, adapts and scales operational decision-making.”

The company’s evolution hints at a shift in the way organizations leverage surveillance systems, using them not only to watch, but also to interpret, learn and assist in ways they previously couldn’t. Lumana’s long-term vision calls for a much tighter integration of video surveillance data, pairing it with insights from sensors, contextual clues and other inputs to enable better real-time situational awareness.

That could one day lead to the development of dynamic crowd control systems in public areas such as stadiums, automated accident response in industrial environments, and support for more effective coordination in emergency situations.

Images: Lumana AI