An apartment complex security system has six layers, not five. Most buying guides skip the one that does the actual work.

The six layers, named

These are the six functional layers of a working apartment complex security system. The first five are present in most properties already. The sixth is the one almost every existing guide leaves out.

Why the sixth layer is the one most installed systems are missing

The buying guides for apartment complex security are written by hardware vendors and aggregators. ButterflyMX writes about intercoms because they sell intercoms. Pelco writes about cameras and recorders because they sell cameras and recorders. The major aggregator sites (SafeHome, Security.org, SafeWise) write affiliate-driven roundups of physical equipment. None of them have a financial reason to recommend a software layer that runs on the recorder a property already owns. So the layer is missing from the published lists, even though it is the layer that decides whether the first five do anything.

The structural failure mode is repeatable. A property has 16 to 40 cameras and a DVR with 7 to 30 days of storage. The cameras record. The DVR stores. The operator finds out about a break-in attempt or a package theft when a tenant complains the next morning. The cameras did exactly what they were sold to do, and the system did not catch the incident. That is layers 3 and 5 working perfectly without layer 6.

The reason the failure is hard to fix from inside the existing five-layer model is that adding a human watcher does not scale. A guard cannot watch 25 feeds at once, costs $3,000 to $5,000 per month per property, and quits. A virtual monitoring service is operator-dependent and adds a vendor in the middle. The architecturally honest fix is a piece of software that watches the feeds 24/7, raises events that match a defined intent profile, and sends them to whoever is on call. That is the sixth layer.

What the sixth layer actually does

An AI analytics overlay is a fairly narrow piece of software. It is not a replacement for the other five layers, and it does not pretend to be a full security platform. The job description fits on a card.

How the overlay attaches to the existing stack

The mechanism that makes the sixth layer cheap is the HDMI composite output of the DVR. Every DVR and NVR built in the last decade paints a 1920x1080 mosaic of all its cameras to a wall monitor in the back office. That signal exists whether anyone is watching it. An overlay device taps that HDMI port, runs one inference pass per frame on the full composite, and maps detected bounding boxes back to per-tile coordinates. At a 4x4 grid each tile is 480x270 pixels; at a 5x5 grid each tile is 384x216. Both are inside the working range of nano-class object detection models and modern vision-language models, which were trained at 224x224 to 416x416 input tensors.

One inference pass covers all 16 to 25 feeds at once. There is no per-camera RTSP credential to recover, no PoE switch to reconfigure, no NVR to replace, no cabling to rerun. The cameras keep painting the wall monitor; the overlay device watches the same wall-monitor signal a guard would have watched. Install is under 30 minutes. The cameras and the recorder do not change at all.

What the cost stack actually looks like

The reason the sixth layer matters in operator terms, not just architecture terms, is the cost gap to the alternatives. A full apartment complex security system replacement (rip out cameras, replace the recorder, run new cabling, license cloud per camera, pay labor) on a 25-camera building runs $50,000 to $100,000 once everything is added up. A full-time night security guard is $3,000 to $5,000 per month per property and watches one camera at a time. A virtual monitoring contract is per-camera per-month and depends on whoever the vendor staffed that night. An overlay on the existing DVR is $450 upfront, $200 per month, and watches every feed at once.

The two paths reach a similar feature set for real-time alerts, intent classification, and natural-language footage search. They are not in the same cost universe. Cross-camera person re-identification at high precision and environmental sensor fusion are the two capability gaps where the replacement path pulls genuinely ahead; for most multifamily incidents (after-hours intrusion, package theft, loitering, vendor no-show) the overlay path is what most operators actually need.

What the first five layers still do (the overlay does not replace them)

One trap to avoid: do not read this guide as an argument that the sixth layer obsoletes the other five. It does not. Access control still decides who is allowed past the door. Intercoms still let the front office identify the person at the gate. Cameras are still the lenses. Alarms and sensors still detect the senses video cannot see (smoke, CO, glass break, water). Recording still gives investigators a 7 to 30 day window of footage to subpoena. Each layer has a job. Removing any of them produces a different failure mode.

The argument is narrower: most multifamily properties in 2026 already have layers 1 through 5 installed and operational. They were installed years ago, they work, and replacing them is not the binding constraint on the property's actual security outcomes. The binding constraint is that the cameras record but no one watches. That is exactly what an analytics overlay layer is for. The honest sequence for an operator with an installed five-layer stack is: keep the five, add the sixth. The honest sequence for an operator with no stack at all is: install layers 1 through 5 from any reputable vendor, then add layer 6 on top.

How to read a vendor pitch through the six-layer lens

When a vendor says "we offer a complete apartment security system," the useful question is which of the six layers they actually do. ButterflyMX is layer 2 (intercom) with a thin layer 1. Verkada is bundled layers 3, 5, and 6, sold only as a package, and only on their cameras. Hikvision and Dahua are layers 3 and 5, with layer 6 sold as separate per-camera AI SoC products that only work on their newest cameras. Hue, Lutron, and Shelly are not security at all; they are deterrence devices that the analytics layer can fire as a webhook on a high-threat event.

An honest "complete system" for a multifamily property in 2026 is almost always assembled across vendors: one vendor for access control, one for the intercom, one for the cameras and recorder (often inherited from a previous owner), one for the alarm, and one for the analytics overlay. The pitch to watch out for is a single vendor claiming all six layers, because the way they reach completeness is to refuse to integrate with the four layers you already own. That is fine for new construction, expensive everywhere else.

What to do this quarter

If your buildings have layers 1 through 5 installed and you are reading this guide because security feels reactive instead of proactive, the work is layer 6. Pick one property, run a 14-day pilot of an analytics overlay on the DVR HDMI output, and measure two things: the rate of alerts that turn out to match a real incident, and the rate of incidents that the operator hears about within 30 seconds instead of the next morning. If those two numbers improve at one property, the case for the rest of the portfolio writes itself.

If your buildings do not have layers 1 through 5 yet, install them from any reputable vendor, in roughly that order, and budget for the sixth layer at the end. The mistake to avoid is buying a vertically integrated stack that bundles all six layers from one vendor, because that path puts you in the rip-and-replace cost universe forever. The architecturally cheap version is five layers of commodity infrastructure with a software layer on top that watches whatever is already there.