Edge AI — run intelligence where the building data is created. LEXI brings lightweight AI inference onto the Universal Building Automation Gateway so building data can be evaluated locally, close to the sensors, equipment and systems generating it. Detect, correlate, forecast and act locally when speed, resilience and building context matter.
NXP eIQ + local inference — turn the gateway into an intelligent building edge. LEXI is using the NXP eIQ AI development environment to deploy and evaluate lightweight AI/ML models on the Universal Building Automation Gateway. The gateway combines local inference with multi-protocol building data rather than acting only as a cloud pass-through. Local inference: run selected AI/ML models directly on the gateway and evaluate results close to the source data. Multi-protocol inputs: use data from wireless sensors, BACnet systems and Modbus equipment in common local intelligence workflows. Rules + AI together: combine model outputs with deterministic rules, thresholds and operating policies instead of relying on AI alone. Managed lifecycle: deploy gateway software and model-related updates through LEXI's managed edge and OTA lifecycle.
Cross-protocol intelligence — understand conditions that no single sensor can explain. The value of Edge AI increases when the gateway can evaluate multiple signals together. LEXI can combine device, equipment, location and time context before deciding whether an event is meaningful. Detection and anomaly identification: identify unusual operating patterns, events and device behavior close to the source. Cross-protocol correlation: relate signals from different sensors and equipment across protocols, locations and operating states. False-positive reduction: use multiple signals, thresholds and model outputs together before escalating an event or triggering a workflow. Enriched events: send higher-value events and contextual results upstream instead of forwarding only raw telemetry.
Lightweight predictive models — move selected prediction closer to building operations. LEXI's current Edge AI work includes deploying and profiling lightweight predictive models appropriate for gateway-class compute and building telemetry. Current model targets: energy consumption prediction, energy demand forecasting, occupancy & occupancy forecasting, equipment predictive failure & maintenance, asset tracking. Energy: evaluate energy consumption and demand patterns closer to connected equipment and meters. Equipment: support model-driven evaluation of equipment condition and remaining useful life where appropriate data is available. Occupancy and presence: use presence, location and related sensor data to support occupancy-aware building intelligence. Model profiling: measure latency, accuracy and gateway resource use so models can be matched to the available edge compute.
Sense, context, reason, act, verify — use AI to inform building action without giving models unrestricted control. LEXI separates AI inference from authorized physical-device actions. Model outputs can inform alerts, workflows and local automation while rules, permissions, thresholds and operating guardrails govern what the system is allowed to do. Sense: collect telemetry and events from connected sensors, meters, rooms and equipment. Context: understand location, asset, time and operating state. Reason: apply AI inference, analytics and deterministic rules. Act: trigger an alert, workflow or approved local control action. Verify: confirm the resulting state and preserve the event for upstream visibility.
Local when local matters — keep time-sensitive intelligence close to the building. Edge AI complements cloud analytics rather than replacing them. Selected workloads run locally when latency, connectivity, resilience or immediate building context make local execution valuable. Lower latency: evaluate local conditions without waiting for every event to travel to a remote cloud and back. Resilient operation: maintain selected inference, rules and workflows when upstream connectivity is interrupted. Less raw-data movement: process and enrich data locally, then forward events, results or selected telemetry upstream. Cloud coordination: use cloud services for broader analytics, dashboards, model and software lifecycle management, and portfolio-level coordination.
Local execution, broader coordination — Edge AI is the local intelligence layer within LEXI's Distributed Intelligence architecture. The Gateway handles selected inference and low-latency response close to the building. Distributed Intelligence connects that local execution with building-wide context, cloud management and portfolio-level coordination. Flow: devices + building systems to the LEXI Universal Building Automation Gateway / Edge AI to building / site coordination to LEXI Cloud and/or existing enterprise systems. Related: Distributed Intelligence, Universal Gateway, Connectivity & Protocols, Managed Edge, White-Label LEXI Cloud.
