Michael Alatortsev, Monitoreal tells Security Journal Americas how AI can reshape physical security.
While I’m relatively new to the traditional video security space, I’ve worked on similar problems – real-time video, privacy and AI – at both startups and big tech.
At Meta, I led privacy and compliance assessments, working with legal and audit teams to reduce AI risk at massive scale.
At Amazon Prime Video, I oversaw product strategy for ultra-low-latency streaming, delivering live events to millions of customers globally.
I entered physical security when I co-founded Monitoreal with a childhood friend, Aydar Yakupov.
For professional monitoring companies, AI fundamentally changes the economics of service delivery. Traditional systems rely on human operators to sift through footage or respond to every motion event – many of which turn out to be irrelevant.
By analyzing video feeds in real time, AI can prioritize only the events that matter – such as a person crossing into a restricted zone, a vehicle loitering after hours or someone going the wrong way.
This means fewer distractions, fewer false dispatches and faster response to real threats. Monitoring teams can handle more properties per shift with greater confidence and less fatigue.
The result? Lower operational costs, higher customer satisfaction and the ability to scale your revenue without needing to scale headcount.
One big misconception is that AI needs the cloud to work. That’s simply not true. It can run entirely at the edge, without sending sensitive data to third parties.
At Monitoreal, we prove that every day – our devices do real-time object detection, filtering and rules processing locally, protecting both performance and privacy.
Another myth is that “AI sees everything.” AI systems are only as good as their data inputs. Detection depends on thoughtful camera placement, optics, lighting and distance.
A third misconception is that any AI system can fully self-calibrate out of the box. The reality is that every customer has different tolerances for false positives or negatives.
A system that’s too sensitive may trigger unnecessary alerts; one that’s too relaxed may miss something important.
That’s why we involve the customer in the initial calibration – so they can decide what settings make sense for their use case and risk tolerance.
We design everything with privacy in mind. Our AI runs locally, eliminating expensive data transfer, storage and inference infrastructure making our solution even more cost-efficient.
Beyond detection, it enables automated security responses – audio, lights and relay actions.
We support both self-monitoring and professional monitoring via platforms like IMMIX and BOLD, and offer a system (GMS) for global device management.
And we’re innovating. Our newest Traffic Analyzer can track hundreds of objects in real time from a single camera view – enabling analysis of pedestrian and vehicle traffic, velocity and behavioral violations.
We’re deploying it at major transportation hubs. While it’s a different product, the core principles remain the same: privacy-centric design, local processing and accurate, low-latency insights.