Industrial AI architecture should begin at the physical process and move outward through controlled layers. This guide outlines the responsibilities of the field, edge, context, intelligence and enterprise planes, and shows where governance belongs.
Layer 1: field and control
PLCs, DCS, instruments, cameras, drives and protection systems remain responsible for deterministic control and safe operation. AI should not create an uncontrolled path around these systems.
Connectivity begins with an explicit inventory of protocols, data owners, update rates and criticality.
Layer 2: edge acquisition and local intelligence
The edge acquires plant data, normalises protocols, buffers during network loss and runs time-sensitive inference. It is also the right place for local data-quality checks, site rules and production-line integration.
SensX and VisionX operate close to the process when sensing, visual inference or local workflow requires predictable response.
Layer 3: operational context
Raw tags become useful when they are connected to assets, process stages, products, work orders, events and operating limits. This layer should expose data quality and uncertainty rather than hide them.
Cortex provides living operational models, while SensX models assets and sensing states across sites.
Layer 4: intelligence and governed action
Advanced analytics and agentic AI use the contextual layer to identify patterns, evaluate scenarios and recommend action. Findings should cite their evidence and pass through human approval where the consequence demands it.
Sentrix is designed around cited, advisory intelligence, and Cortex supports bounded, approved operational steering.
Standards and primary sources used for context.
These links provide technical context. Their inclusion does not claim product certification or compliance unless explicitly stated elsewhere.
- ISA-95 enterprise-control system integrationInternational Society of AutomationModels and terminology for integrating enterprise and control systems.
- OPC UA specificationsOPC FoundationOfficial published OPC UA specifications and companion information models.
- Sparkplug Specification 3.0Eclipse FoundationOfficial Sparkplug specification for MQTT-based industrial data integration.
- ISA/IEC 62443 seriesInternational Society of AutomationCybersecurity requirements and processes for industrial automation and control systems.
