Cyrano isn't a camera. It's the HDMI box that makes the cameras you already own intelligent.

No. Cyrano is a small edge AI device that plugs into the HDMI port on the back of the DVR or NVR that already runs your cameras. It does not replace anything; the cameras keep recording the same way, the DVR keeps writing the same disk, and the box watches what would otherwise have been the wall-monitor signal. The keyword 'cyrano cam' usually comes from someone who heard about the product and assumed it was a camera; it is the layer that makes the cameras you already own send a real-time text the moment something happens.

It is roughly the size of a TV streaming stick or a small set-top accessory. One HDMI input that connects to the DVR's HDMI output, one HDMI passthrough so the back-office monitor keeps working, one ethernet jack, one USB-C power input. It sits on the same shelf as the DVR in the leasing office or building IDF closet. No screen, no buttons. Configuration is done in the management dashboard from any browser.

The DVR is already painting all 25 camera feeds onto a single 1920x1080 HDMI grid for the back-office wall monitor. On a 4x4 grid each camera tile is 480x270 pixels. On a 5x5 grid each tile is 384x216 pixels. Both are inside the working range of nano-class object detection models. One inference pass per frame on the full composite covers every camera in the grid simultaneously, and detected bounding boxes are mapped back to per-tile coordinates so each alert names a specific camera.

Two reasons, both operational. First, per-camera RTSP credentials are typically lost on installs more than five years old; trying to recover them is a multi-week project with the original installer (if they still exist as a business). Second, even when credentials are available, pulling 25 independent RTSP streams across the property network is bandwidth-heavy and forces VLAN changes the IT team does not want to make. The DVR is already doing the multiplexing work on the HDMI port. Tapping that signal is the path of least disruption.

Detection runs entirely on the device. The classifier, the zone logic, and the dwell tracker are all local. The cloud is only used for the management dashboard and to deliver the SMS and phone-call alerts. If the property loses internet, the box keeps detecting and queues alerts locally; when the link comes back, queued alerts ship out in order. The recorded DVR footage is still safe on the DVR's local disk regardless.

Trespassing into prohibited zones (pool deck after hours, rooftop, vacant unit, maintenance closet), loitering at perimeter gates and entry points, package and catalytic-converter theft patterns in parking lots, after-hours intrusion at any zone outside its operating window, and forced-entry signatures at sensitive doors. The output is a single label per event: HIGH THREAT (wakes the on-call manager), LOW THREAT (writes to morning digest with thumbnail), INFORMATIONAL (logs only), or DROP (filtered out before reaching anyone).

$450 one-time hardware cost. $200 per month starting in month two. The hardware payment covers the device, the HDMI capture board, the passthrough cable, and the power supply. The $200 monthly covers the cloud dashboard, the SMS and phone-call alert routing, the natural-language search across recorded footage, and software updates to the on-device classifier. No camera replacement, no IT team required, no infrastructure changes.

Yes. The classifier writes structured event tags to the dashboard for every motion event the DVR fires (subject, zone, dwell, time, threat label, camera tile, thumbnail). Those tags are searchable with natural language queries like 'masked man near gate', 'person loitering at entrance', or 'vehicle in parking lot after 2 AM'. Search returns thumbnails with timestamps and links straight to the DVR's recorded clip for the matching moment.

The box watches the composite signal it gets from the HDMI port, so if a camera tile goes dark or stops updating, the box detects the dark tile and flags it as 'camera offline' in the dashboard with a timestamp. If the DVR itself stops outputting (power loss, firmware crash, disk failure), the HDMI signal disappears entirely and the box raises a 'DVR down' alert. Both failure modes used to be invisible to operators until a tenant complained; the box exposes them inside an hour.

Cyrano de Bergerac. The literary Cyrano stands in the dark and writes the words someone else needs at exactly the right moment. The product does the same job for cameras: it stands behind the recorder, watching what no human is currently watching, and writes the message the on-call manager needs to see at the exact moment that needs seeing. The 'cam' part of the search query is the reasonable mistake people make on first hearing the brand, which is what this page is for.

Yes, in motion and in price. Verkada and Rhombus require ripping out the existing cameras and replacing them with their cloud-connected hardware, which runs $10,000 to $25,000 in camera and labor cost per property plus ongoing per-camera subscriptions. They are built for enterprise customers in a single rip-and-replace motion. Cyrano sits on top of the cameras already at the property; the cameras and the recorder do not change. For a Class B or Class C multifamily portfolio where rip-and-replace is not on the table, Cyrano is the layer that adds the alerting without the overhaul.

There is a 30-minute demo call on the booking link with the founder. Bring your camera count and DVR brand. We walk through the dashboard live with the Fort Worth deployment as a reference (20 incidents caught in 30 days at a Class C multifamily property, including a break-in attempt, customer renewed after the first month). If the architecture is a fit, install ships in the following week.