Integrating Physical AI and IoT into modern manufacturing
Combining Physical AI, the Internet of Things (IoT), and edge platforms creates systems capable of responding to physical processes in real time. However, successful deployment of these technologies requires overcoming data fragmentation.
According to Gartner forecasts, by 2028, more than half of enterprise generative AI models will be domain-specific. This requires high-quality data, whereas in most manufacturing plants, information is currently scattered across different systems in incompatible formats.
The OT/IT gap and how to bridge it
Traditionally, enterprises have two isolated environments: operational technology (OT, such as SCADA, PLC, MES) and information technology (IT, including ERP and CRM). They use different protocols and architectures, which prevents the creation of a unified picture for AI.
Instead of trying to clean all data at once, experts recommend an iterative approach. It is best to identify priority use cases, such as predictive maintenance. To achieve this, real-time equipment status data (SCADA), execution history (MES), and maintenance schedules (ERP) are integrated on a local edge platform.
The role of Edge AI, 5G, and risk management
Processing data directly on edge platforms minimizes latency, which is critical for robotics. According to the Ericsson Mobility Report, by the end of 2027, 5G technology will become dominant in mobile access, providing the necessary speed for IoT systems.
Since Physical AI directly controls physical equipment, errors or cyberattacks can lead to accidents. To minimize threats, organizations use the NIST AI RMF 1.0 framework and ISA/IEC 62443 industrial cybersecurity standards. A systematic approach to data readiness and security is crucial for manufacturing automation.
The effect on the market
The shift toward domain-specific AI models by 2028 means manufacturers must resolve data fragmentation or risk falling behind in automation. Furthermore, because Physical AI directly controls machinery, failing to secure these systems and bridge the OT/IT gap poses direct operational and physical safety risks to enterprises.
Recommendations
- Use an iterative approach: Avoid trying to clean all data at once. Instead, identify priority use cases such as predictive maintenance.
- Integrate key systems: Combine real-time equipment status (SCADA), execution history (MES), and maintenance schedules (ERP) on a local edge platform.
- Secure the infrastructure: Apply the NIST AI RMF 1.0 framework and ISA/IEC 62443 industrial cybersecurity standards to minimize the risk of accidents and cyberattacks.
Prepared by a Software Ukraine member. Original publication.