A building that tells you what’s about to happen
Most building systems tell you what already went wrong. xSERVA’s AI and IoT layer is built to get ahead of that — learning the normal behavior of every connected system and flagging the moment something starts to drift, before it becomes a fault, a comfort complaint, or a wasted kilowatt.
Book a DemoPredictive maintenance
xSERVA’s models learn the operating signature of each piece of equipment — the vibration, runtime, cycling frequency, and load pattern that count as normal for that specific AHU, chiller, or pump. When a unit starts to deviate from its own baseline, the platform flags it as a developing issue, weeks before it would trigger a hard alarm on its own. Maintenance teams get a prioritized list of equipment trending toward failure, with the specific parameter driving the flag, instead of reacting to breakdowns after the fact.
Anomaly detection
Beyond individual equipment, xSERVA watches for patterns that don’t fit — a zone drawing power outside its usual profile, a sensor reporting values inconsistent with its neighbors, a schedule override that keeps recurring. These anomalies often precede the alarms that traditional threshold-based systems eventually catch, giving operators a earlier, quieter signal to investigate before it escalates into something visible to occupants.
Ask your building a question
xSERVA includes a natural-language query interface so operators don’t need to build a custom report to get an answer. Type a question the way you’d ask a colleague — “Ask xSERVA: which AHUs are trending toward failure this month?” — and the platform returns a direct answer drawn from live and historical data, not a dashboard you have to interpret yourself. The same interface handles questions about energy consumption, alarm history, or equipment status across one site or an entire portfolio.
Digital twin visualization
A 3D digital twin of the building renders live point data spatially — zone temperatures, equipment status, and alarms mapped onto the actual floor plan rather than a flat list of tags. Operators navigate the building visually to understand where an issue is physically located, which is often faster than reading a point name and translating it to a location in your head.
Edge computing
For latency-sensitive control loops and sites with unreliable connectivity, inference can run at the edge — on local hardware inside the building — so anomaly detection and predictive alerts keep working even if the link back to a central instance drops. Edge and cloud processing share the same models, so insight quality doesn’t degrade depending on where it runs.
IoT sensor fusion
Modern buildings increasingly layer lightweight IoT sensors — occupancy, air quality, vibration, humidity — on top of traditional BACnet, KNX, and Modbus infrastructure. xSERVA fuses this MQTT-borne sensor data with existing controller and meter data in the same model the AI layer already reads from, so a new occupancy sensor immediately makes every prediction and anomaly check that touches its zone more accurate, without a separate integration project.
Intelligence, grounded in unified protocol data
None of this works as a bolt-on. The AI/Analytics layer sits directly above the unified protocol data that BACnet, KNX, MQTT, and Modbus devices already feed into xSERVA, which is what lets predictions span every subsystem in a building rather than just the one protocol a point-solution happened to support. See how that underlying architecture is structured.
Frequently asked questions
What kind of AI does xSERVA use?
xSERVA applies machine learning models trained on live building data for predictive maintenance, anomaly detection, and natural-language querying, running continuously over the same data that powers monitoring and control.
Does the AI layer need extra hardware?
No new field hardware is required. The AI layer runs on top of data already flowing through the protocol layer from existing BACnet, KNX, MQTT, and Modbus devices.
Can I ask xSERVA questions in plain language?
Yes. The natural-language query interface lets operators ask questions like which AHUs are trending toward failure this month and get a direct answer drawn from live and historical data.
How does the AI layer relate to the rest of the platform?
AI and IoT intelligence sit as a layer above the unified protocol data described on the platform architecture page, so every insight is grounded in the same live data as monitoring and alarms.