Agentic AI in manufacturing requires unified, low-latency data access across ERP, MES, and historian systems. According to Peter Sorowka of Cybus GmbH and Sebastian Trolli of Frost & Sullivan, data silos, missing metadata, and inconsistent architectures are the primary barriers stopping autonomous AI agents from acting reliably on the factory floor.
Should Agentic AI Use a Central Repository or Direct Access?
For agentic AI in manufacturing, both central repositories and direct system access are valid approaches, but neither alone is sufficient. Peter Sorowka, CEO of Cybus GmbH, states that agentic AI needs unified access, low latency, and consistent data models across ERP, MES, and historian systems. Static data lakes are not enough.
Peter Sorowka, Cybus GmbH: Both approaches exist, but agentic AI needs more than static data lakes. Think of it less as a parking lot and more as a real-time data highway. Unified access, low latency, and consistent data models are non-negotiable.
In greenfield industries like battery manufacturing, new equipment is now expected to connect not just to power but also to a unified namespace from day one. For legacy plants, integration and standardization become the harder — but essential — first steps.
Breaking Down OT/IT Silos
Resolving OT/IT data silos for agentic AI requires both technical frameworks and organizational change. Peter Sorowka of Cybus GmbH identifies unified namespace and Model Context Protocol (MCP) as the most effective architectures. Sebastian Trolli of Frost & Sullivan adds that real-time pipelines must be paired with metadata so agents understand what data actually represents.
Peter Sorowka, Cybus GmbH: Unified namespace and Model Context Protocol (MCP) are among the most effective frameworks. But the bigger hurdle is organizational: if procurement doesn’t require data-friendly interfaces when buying machines, integration later becomes a nightmare.
Sebastian Trolli, Frost & Sullivan: Even when silos are bridged, context is critical. Real-time pipelines alone are not enough. Without metadata and models that explain what the data represents, agents cannot act reliably.
How Does Agentic AI Map to ISA95 and ISA88?
Agentic AI does not replace ISA95 or ISA88 standards. It extends them. According to Sebastian Trolli, Head of Industrial Automation & Software at Frost & Sullivan, agentic AI acts as a cognitive layer across the ISA95 pyramid, with Level 3 (MES) identified as the primary “sweet spot” for orchestrating workflows, scheduling tasks, and resolving exceptions.
Sebastian Trolli, Frost & Sullivan: There’s no single blueprint yet, but ISA95 provides a useful starting point. Agentic AI acts as a cognitive layer across the pyramid:
- Levels 0–1 (sensors, PLCs): Monitoring and advisory roles, not direct control.
- Level 2 (SCADA/HMI): Agents aggregate views across HMIs, assist operators, and exchange commands.
- Level 3 (MES): The “sweet spot” where agents orchestrate workflows, schedule tasks, and resolve exceptions.
- Level 4 (ERP, supply chain): Aligning enterprise goals with shop-floor execution.
Rather than replacing ISA95 or ISA88, agentic AI extends them, linking operational layers to business objectives with real-time intelligence.
Why Do Agentic AI Systems Need Data Ecosystems?
Factories that want agentic AI to succeed need more than large stores of historical data. They need living, connected data ecosystems where agents can access clean, contextual, and current information. That means unified namespaces, metadata that gives meaning to data streams, and procurement strategies that treat interoperability as a firm requirement from day one.
The lesson is simple: factories don’t need more data lakes — they need living, connected data ecosystems where agents can think and act with confidence.
Speakers contributing to these answers:
- Peter Sorowka, CEO, Cybus GmbH
- Sebastian Trolli, Head of Industrial Automation & Software, Frost & Sullivan
Related articles:
- Agentic AI and IIoT: Complex Questions and Expert Perspectives
- Agentic AI in manufacturing does not need to begin with a moonshot
FAQ
1. What data architecture does agentic AI require in manufacturing?
Agentic AI in manufacturing requires unified access to data from systems such as ERP, MES, and historians, with low latency and consistent data models. Peter Sorowka of Cybus GmbH describes this as a real-time data highway rather than a static data lake. Unified namespaces and metadata that explain what data represents are both essential for agents to act reliably.
2. What is a unified namespace in industrial manufacturing?
A unified namespace is an architecture that gives all systems and applications a single, consistent point of access to manufacturing data. Peter Sorowka of Cybus GmbH identifies it as one of the most effective frameworks for resolving OT/IT data silos. In greenfield facilities such as battery plants, new equipment is now expected to connect to a unified namespace from the start of commissioning.
3.How does agentic AI fit into the ISA95 automation pyramid?
Agentic AI acts as a cognitive layer across all levels of the ISA95 pyramid rather than replacing it. According to Sebastian Trolli of Frost & Sullivan, Level 3 (MES) is the primary area where agents add value by orchestrating workflows and resolving exceptions. At Levels 0 and 1, agents take monitoring and advisory roles only, without direct control of sensors or PLCs.
4. Why is metadata important for agentic AI in manufacturing?
Metadata tells an AI agent what a data stream actually represents, including its raw value. Sebastian Trolli of Frost & Sullivan explains that real-time pipelines alone are not sufficient; without metadata and models that provide context, agents cannot act reliably. Bridging OT/IT silos technically is only part of the solution; the data must also carry meaning for agents to make correct decisions.
Related Reading
The Human Costs, Regulatory Compliance, and the Workforce Transformation Playbook (Part 3 of 4)
