Most integration failures in smart building automation are not caused by the automation logic. They are caused by the data feeding it. A schedule-based HVAC controller running a "meeting room" profile for an empty room is technically working perfectly — it just has no idea the room is empty. That gap between assumption and reality is where energy and money leak, and it is the first thing worth fixing before you write a single automation rule.
Automation partners often build the rules first and bolt sensors on afterwards. Reverse that. Your building automation system can only act as well as it can sense. Real-time occupancy — the actual number of people in a zone right now, not a badge swipe from three hours ago — is the input that lets you turn "always-on" behaviour off without guessing. Vemco has worked on this since 2005, and the pattern holds across offices, universities and public buildings: the sensing quality dictates the automation ceiling.
When you evaluate sensors, ask about accuracy honestly. AI-based counting typically reaches 98–99% under good conditions — even lighting, clean sightlines, predictable visitor flow — with a contractual minimum of 96%. Anyone promising a flat 99% guarantee regardless of layout is selling you a number, not a measurement. Lighting, store or floor layout, and how people actually move through a space all move the figure. Plan for the 96% floor and be pleased when conditions push you higher.
Three systems consume occupancy data directly, and each has different tolerances for latency and reliability:
This is exactly the routing that VemFusion handles: it takes occupancy data and connects it to HVAC, BMS and security, so each system receives what it needs in the form it expects. Meanwhile VemSpace works on the analytics side — space utilisation and facility optimisation — answering the slower, strategic questions about which floors are underused and which meeting rooms are booked but empty.
Real-time occupancy with alerts at a predefined limit is one of the most useful features to configure early, and one of the most commonly rushed. A predefined limit is a business decision, not a technical default. A lecture hall's cap is a fire-safety figure. A café's cap might be a comfort or service figure. Get facility managers and safety officers in the room to agree these numbers before commissioning, because retrofitting thresholds after users have grown used to false or missing alerts erodes trust in the whole system.
A practitioner note: build a deadband into your threshold logic. If your limit is 50 and the count oscillates between 49 and 51 as people cluster near a doorway, a naive rule will fire and clear an alert repeatedly. Trigger at 50, clear only when the count drops to, say, 45. That single hysteresis rule prevents the notification storms that make staff mute alerts entirely.
In offices and public buildings, counting everyone equally distorts the picture. Staff who are present all day inflate occupancy figures and skew utilisation analytics. Sensors that support staff exclusion keep visitor and occupant data clean, which matters both for accuracy and for the privacy conversations you will have with works councils and data-protection officers. Treat this as a design decision at the sensor level, not a filter you apply later in a report — it is far cleaner to exclude at source.
For teams planning a rollout, this order reduces rework:
When the data layer is right, the automation becomes quiet. HVAC ramps down in empty wings without anyone noticing. Capacity alerts fire only when they matter. Facility managers stop debating whether a floor is used and start reading it. The measure of a good smart building automation project is not how many rules you wrote — it is how much of the building now takes care of itself without a person overriding it every morning.
If you are scoping an integration and want occupancy data that stands up to HVAC, BMS and security demands, talk to the Vemco team about how VemFusion and VemSpace fit your building. Get in touch with Vemco here.