Industrial IoT connects machines, sensors, controllers and plants to software, so that production data becomes accessible, contextualized and actionable from the edge to the cloud. This page collects my work on industrial IoT architecture: the independent landscape, the architecture guides, use cases across manufacturing, automotive, energy, transportation and logistics, and the talks. It covers the plant floor and everything that connects it to the enterprise, including data streaming, data integration, workflow orchestration and agentic AI.

Why the architecture decides what industrial IoT delivers

Industry 4.0 was announced as a revolution more than a decade ago. Most plants stayed in pilot mode, with point-to-point integrations and proprietary protocols. The market answered with architecture. The hyperscalers retreated to infrastructure primitives, and OT-native platforms, open standards and open source filled the application layer. Three decisions now shape every industrial IoT program: which open standards carry the data, how plant data is contextualized and distributed, and where the plant connects to the enterprise.

A Unified Namespace (UNS) is an event-driven architecture pattern in which all plant data is published into a centrally structured, hierarchical namespace, typically an MQTT broker organized along ISA-95, so that every application consumes current, contextualized data from one place. Producers publish once, consumers subscribe independently, and each additional application costs no more than the last. The UNS replaced point-to-point integration as the dominant pattern inside the plant. It is the same shift that event streaming brought to enterprise IT, which is why the two worlds connect so cleanly at the boundary.

OPC UA and MQTT are complementary standards that work at different layers, with OPC UA normalizing device-level semantics at the machine and MQTT distributing contextualized events across the plant and to the enterprise. Sparkplug B makes MQTT payloads interoperable across vendors. The practical pattern of 2026 is OPC UA at the equipment interface, contextualization in an industrial DataOps layer, and MQTT with Sparkplug B as the distribution fabric. A vendor that forces a choice between the two is describing the limits of its own product.

The IT/OT boundary where the plant meets the enterprise

The IT/OT boundary is the architectural line where plant systems hand contextualized operational data to enterprise platforms and receive orders, models and coordinated decisions in return, with each OT layer paired to a counterpart on the IT side. Industrial IoT reaches beyond the plant gate and relies on the enterprise data stack on the other side. The Unified Namespace exchanges real-time events with data streaming, where Apache Kafka and Apache Flink carry plant events into the enterprise and back. Applications such as MES exchange orders and transactions with ERP and business systems. The intelligence layer exchanges models and training data with the lakehouse. Workflow orchestration decides and coordinates above the data movement, with AI agents as governed steps and a person approving every action that reaches the control layer. Closing the loop into control logic stays with the PLCs, virtual PLCs and SCADA systems, because determinism and safety certification draw that line.

This is how the topics on this site fit together. The Industrial IoT Landscape covers the plant side. The Data Streaming Landscape and the Data Integration Landscape cover movement and connection on the IT side. The Trusted Agentic AI Landscape covers the agents that act on plant data, and the Process Intelligence Landscape covers the coordination layer above all of it, including workflow orchestration.

The Industrial IoT Landscape

The map below shows the four categories of the landscape from edge to cloud, from industrial connectivity to applications and industrial AI.