How Foresight works: four layers of intelligence
Foresight doesn't replace your BMS — it transforms it into an intelligent system. Four layers sit on top of your existing infrastructure, turning raw data into prioritised actions in days.
The intelligence stack
Cognitive Layer
Natural language, prioritised actions
AI & Machine Learning
Anomaly detection, predictions
Expertise Engine
Codified engineering rules
Knowledge Graph
Semantic building model
Your existing BMS
Foundation — unchanged
Your existing BMS
Foresight starts with what you already have — whatever vendor, protocol, or age. No rip and replace. We connect via edge gateway through your existing network. BACnet, Modbus, OPC-UA, and proprietary systems are all supported.
Compatible with
Piscada BAS · Schneider Electric · Siemens · Honeywell · Johnson Controls · Trend · Distech · any BMS with standard protocols
Timeline: hours to connect, data flowing same day.
Knowledge graph
Raw data points don't mean much without context. Layer 1 maps every asset and relationship in your building using Brick Schema — an open standard for building metadata.
The graph knows that “AHU-3-SA-TEMP” isn't just a number — it's the supply air temperature sensor in Air Handling Unit 3, serving Floor 2 West, controlled by valve V-204, impacting zones R-201 through R-215.
Portfolio-wide queries. "Show me all AHUs running outside scheduled hours" — across 1,500+ buildings instantly.
Root cause tracing. Follow system relationships automatically to find why a problem is happening.
Impact prediction. Know which zones are affected before tenants complain.
Timeline: 1–2 days to map a building. Same model scales to 1,500+.
Expertise engine
This is where faults get caught. Layer 2 contains executable rules built since 2010 — the accumulated knowledge of building engineers who've diagnosed thousands of issues across 1,500+ buildings. Every rule is traceable and explainable, never a black box.
Example rules
Heating valve >10% open while outdoor temp >15°C for 2+ hours → potential heating/cooling conflict
AHU at 80%+ capacity during scheduled "off" hours → check for override or schedule issue
Room temp deviation >2°C from setpoint for 4+ hours → diagnose valve, damper, or control loop
Timeline: active immediately, first findings within 48 hours.
AI & machine learning
Layer 3 detects patterns invisible to human engineers and even to rule-based systems. Models trained on millions of data points across the portfolio catch anomalies, predict failures, and optimize operations continuously.
Anomaly detection
Learns normal behaviour per building and system. Flags deviations that signal degradation before any rule fires.
Predictive maintenance
Models equipment degradation curves. Predicts failures based on usage patterns and similar equipment across the portfolio.
Energy optimisation
Identifies waste during holidays and off-hours. Recommends setpoint changes that maintain comfort while cutting cost.
Timeline: active from day one, accuracy improves over the first 30 days.
Cognitive layer
The top layer translates everything below into plain language. Findings become prioritised actions. You can ask questions in natural language and get answers that draw on all four layers — with full traceability back to source.
Example daily report
Building 4A: AHU-3 supply temp drifting
Heating valve V-204 stuck at 45% open. Comfort impact in 24h. Energy waste: €45/day.
Building 7B: Holiday mode still active
Heating schedule not restored after Christmas. 12 zones affected.
Building 2C: Energy optimisation opportunity
Reduce heating setpoint 0.5°C during 6–8am. Potential saving: €120/month.
Conversational AI
You ask
“Why is energy consumption high in Building 4 this week?”
Foresight responds
“Three AHUs ran 18 hours beyond schedule due to manual overrides not reset after maintenance. Detected Jan 15th. Cost impact: €340 this week. Want me to reset the schedules?”
Learn more: Explore Foresight AI →
Putting it all together
Connect & map
Your BMS streams data. Knowledge Graph maps every asset and relationship. Hours to connect, 1–2 days to map.
Detect & learn
Expertise Engine applies rules, AI models detect anomalies. First findings within 48 hours.
Prioritise & act
Cognitive Layer ranks by impact. Daily report every morning. Conversational AI ready for your questions.
Intelligence in days, not months
Most platforms take 3–6 months to configure. Foresight delivers first insights within 48 hours of activation — no army of consultants required.
See it working
on your buildings.
Book a technical demo and we'll walk through the four-layer architecture with your actual buildings and infrastructure.