EDITOR’S CHOICE


constraints apply. Without this linkage, analytics must rely on inference. With MES, context is explicit. MES also acts as a dramatic accelerator for


establishing a Common Data Model. Because MES is already structured around core manufacturing concepts, assets, processes, materials and execution states, it provides a practical blueprint for data consistency. Rather than defining a Common Data Model from scratch, organisations can align it to the MES structure they already use to run operations. This alignment reduces ambiguity, shortens implementation timelines and ensures that the data model reflects how the factory actually works rather than an abstract ideal.


FROM EXECUTION SYSTEM TO REAL TIME INTELLIGENCE PLATFORM As manufacturing evolves, MES is no longer sufficient as a standalone execution layer. It is increasingly extended into a broader data and intelligence platform. This evolution allows MES to support real time operational decisions while also enabling advanced analytics, machine learning and longer-term optimisation. Real time insights allow teams to respond


immediately to deviations in quality, performance, or energy consumption. Historical and contextualised data supports trend analysis, predictive maintenance, root cause investigation and continuous improvement initiatives. Machine learning models trained on this consistent foundation can detect subtle patterns that would otherwise go unnoticed and can generalise those insights across lines and facilities. By combining MES with a unified data platform,


factories break down traditional silos. Decisions are no longer made independently by production, maintenance, or quality teams. Instead, they are informed by a shared operational view, allowing the factory to optimise as a system rather than a collection of parts.


manufacturing data without a strong foundation. Contextualisation through MES and a Common


Data Model provides that foundation. It ensures that AI systems understand what data represents, how different elements relate to one another and which constraints apply. With this grounding, AI moves from speculative reasoning to operational relevance, supporting decisions that align with how the factory actually runs.


HUMAN IN THE LOOP AND THE IMPORTANCE OF EXPLAINABILITY Even with advanced intelligence, manufacturing remains a human-driven discipline. Decisions often involve tradeoffs between competing objectives such as throughput, quality, cost and energy. These tradeoffs require judgment, accountability and experience. Human in the loop validation ensures that AI


recommendations are reviewed and understood before action is taken. This makes explainability essential. Operators and engineers need visibility into why a recommendation was made, which data influenced it and what outcomes are expected. Explainable models build trust, reduce resistance to adoption and enable teams to learn from both successful and unsuccessful outcomes. This collaborative relationship between humans


and intelligent systems turns AI into a force multiplier rather than a black box. Over time, feedback from human validation improves data quality, refines models and strengthens decision making across the organisation.


AI NEEDS CONTEXT AND CONSISTENCY TO BE TRUSTED The rise of AI, particularly large language models, introduces powerful new capabilities for reasoning, summarisation and interaction. However, these models are highly sensitive to the quality and structure of the information they consume. Without proper context and consistency, they can generate confident but incorrect outputs, a phenomenon often described as hallucination. In a manufacturing environment, hallucination


is not an abstract risk. An AI system that misunderstands operational context may recommend actions that conflict with safety constraints, quality requirements, or production realities. This is why AI cannot be layered on top of


MCP AND THE FOUNDATION FOR INTELLIGENT ACTION As AI systems move from insight to action, they require a standardised way to interact with manufacturing data and systems. This is where the Model Context Protocol plays a critical role. MCP provides a structured interface through which AI models can access context, retrieve information and invoke tools in a governed and predictable manner. In practical terms, MCP defines how an AI


system understands what data is available, how it is structured and what actions it is allowed to take. It separates the intelligence of the model from the operational systems it interacts with, ensuring that access is controlled, auditable and consistent. For manufacturing, this is essential. It allows AI to operate safely within complex environments without hard-coding integrations or bypassing existing controls. MCP enables tools that allow AI to query MES


data, retrieve historical performance, request quality records, or initiate workflows. These tools do not grant unrestricted access. They operate within defined boundaries, ensuring that actions


AGENTS AND THE MOVE TOWARD ADAPTIVE MANUFACTURING The next evolution of AI in manufacturing comes through agents. Unlike traditional analytics or monolithic AI systems, agents are designed around specific goals and limited scopes. The most effective agents focus on narrow domains, such as monitoring a specific process, optimising a single KPI, or managing a defined decision loop. They typically use a small number of tools, sometimes even a single tool, which makes their behavior easier to understand, validate and control. Rather than building one agent to manage


everything, complex challenges are addressed through collaboration between multiple specialised agents. Each agent contributes within its domain, and larger workflows emerge from their interaction. Standards such as the A2A protocol enable this coordination, allowing agents to communicate, share context and delegate tasks in a structured way. This multi agent approach mirrors how


factories already operate, with specialised roles working together toward shared objectives. When combined with MCP, it allows agents to act within clearly defined boundaries while still contributing to broader operational goals. As these agent-based systems mature,


factories begin to shift from reactive management to continuous adaptation. Agents can monitor conditions, correlate signals across systems, propose actions and coordinate responses at machine speed. Humans remain in control, setting objectives, validating outcomes and handling exceptions, while agents manage complexity that would otherwise overwhelm manual processes.


are transparent and subject to governance.


In this future, manufacturing systems become


living environments that sense, learn and adjust continuously. The factories that succeed will not be those with the most data, but those with the clearest context, the strongest foundations and the intelligence to turn understanding into action faster than their competitors.


Critical Manufacturing www.criticalmanufacturing.com MARCH 2026 | FACTORY&HANDLINGSOLUTIONS 25


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40