OPC Foundation Advances OPC UA for the AI Era with Companion Specifications Optimized for Agentic AI
The OPC Foundation today announced the next step in its OPC UA for AI activities: building on the initial prototyping work published in the UA-for-AI-Prototype GitHub repository, the OPC Foundation will extend this work to convert all (over 430) OPC UA Companion Specifications into formats optimized for retrieval-augmented generation (RAG), model context protocol (MCP) and AI-assisted engineering workflows.
The prototype repository already demonstrates how OPC UA core specifications have been transformed into AI-ready assets such as Markdown files, image descriptions, token-optimized RAG chunks, vector embeddings, and interfaces for MCP and REST-based querying.
OPC UA Companion Specifications are domain-specific information models built on OPC UA that define how specific industries, devices, and use-cases expose information in a consistent, machine-readable way. They typically include both a written specification and a UA NodeSet file for implementation. The OPC UA Companion Specifications are created by OPC Foundation working groups, joint working groups with partner organizations, or by external organizations adopting OPC UA modelling practices.
The benefit of Companion Specifications is that they enable semantic interoperability across vendors, products, and sectors. By standardizing not just connectivity but also the meaning and structure of data, Companion Specifications allow applications, and digital solutions to discover, understand and integrate industrial information more easily, without requiring repeated effort to create correct mappings for every vendor implementation. OPC UA combined with these industry standard information models, enables interoperability at the semantic level. With the rise of industrial AI, engineering copilots, and natural-language interfaces to operational technology, the value of these standardized information models extends even further.
Building on that prototype work, the OPC Foundation intends to broaden the scope from the initial OPC UA specification set to the wider ecosystem of OPC UA Companion Specifications, making them more accessible to AI systems in a consistent, scalable, and trustworthy way. The goal is to help developers, system integrators, equipment manufacturers, and end users leverage Companion Specification content more effectively with AI-based systems in contextual engineering assistance, semantic search, automated documentation support, onboarding acceleration, and agentic AI. The OPC Foundation already makes Companion Specifications and NodeSets accessible through resources such as the UA-NodeSet repository and the UA Cloud Library, providing an established foundation for broader machine-readable and searchable access.
By making Companion Specifications RAG-optimized, the OPC Foundation aims to reduce friction by integrating sound industrial knowledge in modern AI systems while preserving the rigor, governance, and interoperability principles that have made OPC UA the leading standard for industrial information exchange. This effort reflects the OPC Foundation’s commitment to ensuring that OPC UA information models remain not only interoperable among machines and software systems, but also understandable and actionable for the next generation of AI-powered tools.
“Agentic AI systems present a unique opportunity for industrial automation, but to be successful, they need to provide correctness and reliability as good or better than existing engineering approaches. Grounding them in the established semantic interoperability of OPC UA is an important step towards reliability and acceptance” said Dr. Holger Kenn, leader of the OPC Foundation AI working group.
“OPC UA Companion Specifications already provide the semantic backbone for interoperable industrial data exchange,” said Stefan Hoppe, President OPC Foundation. “By extending our AI work to these specifications, we are helping ensure that the rich domain knowledge created by the OPC Foundation and its partners can be used more effectively by engineers, developers, and intelligent applications.”
News release from OPC Foundation, 20/04/2026