When you install Watchmark, your building suddenly becomes a smart building that "knows" whether it's occupied by polling images from your security cameras and running AI human detection. To do this, your security cameras need to be capable of exposing a URL or RTSP stream that the access panel can reach over your local network.
To handle cases like public lobbies, insecure spaces, etc, each building can have multiple security zones, and a single camera can even have its field of view split so it detects people in multiple security zones (inside a glass door vs outside the door, for example).
Watchmark is a new kind of security system - no more remembering to arm the alarm when you leave, dealing with users forgetting the pin, or waking up in the middle of the night with a vague "motion detected" alert. When someone who is allowed to access a building opens a door, the building automatically knows that it's occupied for a certain amount of time (30 minutes is a good default in many cases). As long as the person continues to be detected, the building occupancy will automatically extend for as long as they're in the building, and when they leave the building will realize that it's now unoccupied.
If a human is detected in a building that thinks it should be unoccupied, an alarm will be sent to any number of users who are on-call at that time - Watchmark allows for multiple on-call rotations with automatic escalation until each incident is resolved.
The real world is complicated, so Watchmark uses flexible rules to detect and alarm - for example, when someone from the Cleaning Group lets themselves into the building, it might consider itself occupied for 3 hours instead of the usual 30 minutes, or you might set up a group called Night Owls to avoid false positives for employees who typically work late. Some organizations choose to only trigger certain types of alarms late at night or on holidays.
The alarm that's sent to the on-call responder includes the snapshot from the camera that triggered the alarm, so they can immediately classify it as either a real threat or a false positive and take appropriate action (log in to the camera system to get a better picture, call law enforcement or internal security, etc).
Of course, AI human detection isn't perfect and there will always be some false positives, but our models are constantly improving and increasing your ability to only get notified during incidents you care about!
Other sorts of alarms can also be triggered, including panic buttons being pressed or doors jammed open and failing to close. Your buildings will almost seem self-aware, and you'll wonder how you ever managed to get by without smart buildings!