Security cameras have always been good at recording and bad at understanding. Eluviant wants to close that gap. The AI company formerly known as IntelexVision on Tuesday launched Aurora Flow, a frontier “video understanding” model built for live, enterprise-scale surveillance — and used the same announcement to unveil a full rebrand under its new name.
The pitch is that Aurora Flow moves past the object-in-a-frame logic of traditional video analytics. Instead of flagging that a person or vehicle appears, the model is designed to read how a sequence of movement plays out over time and then judge whether a human actually needs to look. Eluviant says it is already running in live deployments, can operate fully air-gapped, and works across multiple cameras in near real-time — the kind of constraints that matter to operators in critical infrastructure and government who cannot pipe footage to the cloud.
Why it matters
Control rooms drown in alerts. The headline stat Eluviant is leaning on: in one deployment, Aurora Flow collapsed roughly 4,000 potential events in a day down to just seven verified alerts — about 0.2% of the raw volume. Behaviors that have historically needed human eyes to catch reliably, such as equipment tampering, unsafe climbing, dangerous driving, fighting and theft, are exactly the messy, time-dependent events the company says its model is built to recognize as they happen.
Aurora Flow is not a standalone leap so much as an extension of a stack Eluviant has been shipping for years: an unsupervised self-learning engine that surfaces genuinely unforeseen events, plus a vision-language model, Aurora, that the company says has sat inside the live alerting decision for the past 18 months.
“We believe Aurora Flow is a frontier AI model in surveillance and a step change in what video intelligence can deliver, moving beyond detection and into genuine understanding and evaluation of behaviours and actions in complex live environments,” said Rafik Lamri, Eluviant’s regional director for the Middle East, Turkey and Africa. “Things like fighting, climbing and theft have typically required human eyes to detect them accurately — now we can help operators focus on what needs their urgent attention by putting AI into the alert decision.”
Founded in 2017, the company is betting that existing camera networks are underused assets rather than just security tools. It cites an enterprise video-intelligence market it expects to reach $30 billion by the end of the decade, with the Middle East a hot spot: the UAE and Saudi Arabia lead a regional video-surveillance market valued at around $4.3 billion, driven by smart-city projects and government-mandated security. Eluviant says it now has more than 250 deployments across five continents and 60-plus partners, with customers including Airbus, DP World, Prosegur and Vodafone.
The skeptic’s note: “understanding” is a heavy word for a system that is ultimately pattern-matching movement, and the eye-catching reduction figures are vendor-supplied from a single site. AI that infers intent and behavior from live footage also lands squarely in the middle of the surveillance-and-privacy debate, where accuracy, bias and oversight are legitimate concerns — especially in the “most secure and sensitive environments” Eluviant is targeting. Still, the direction is clear: the industry is pushing video analytics from spotting what is in a frame toward interpreting what is happening in it.
