From Digital Thread to Agentic Operations: What Manufacturing Leaders Need Next

Manufacturing leaders have spent the past decade digitizing individual functions with the enterprise. Engineering teams modernized PLM. Plants invested in automation, sensors, robotics, and MES. Supply chains built planning systems and control towers. Quality teams adopted predictive quality control & analytics. Service organizations began using connected product data. These investments improved local performance, but they also created a new leadership problem: the enterprise may have become more digital, yet it’s still not more coherent.

That distinction is increasingly critical in today’s manufacturing environment. Our global manufacturing clients now operate with a level of product and operational complexity that was unimaginable a generation ago. This complexity propagates rapidly across functions and at speed. Products are more configurable, software-defined, connected, and tightly regulated. A single design change can ripple across supplier readiness, manufacturing processes, software configurations, field diagnostics, sustainability reporting, warranty exposure, and regulatory compliance. Similarly, a quality issue may originate as a component deviation, extend into production, surface in field performance, and ultimately impact customer trust. Even localized supply disruptions cascades across geographies and impact capacity, working capital, customer commitments, and service levels.

This is why the digital engineering conversation has moved beyond tooling. The question is no longer, “Do we have digital systems?” Most large manufacturers already do. The sharper question is, “Can those systems create a trusted operating foundation for decisions across the product lifecycle?”

Fig 1: What Changes for the CXO?

Why the Digital Thread Is Becoming the Enterprise Nervous System

That foundation is what the industry often calls the digital thread.

At its simplest, a digital thread is the connected flow of product and process information across the lifecycle, from concept, design, and engineering to manufacturing, quality, service, and eventual retirement. It links the “why” of a product, such as customer requirements and engineering intent, with the “what” of product structures and configurations, the “how” of manufacturing processes, and the “how well” of quality and field performance.

In our customer ecosystem, where a digital thread has been established, leaders are able to understand how decisions in one part of the lifecycle impact cost, quality, compliance, delivery, sustainability, and customer experience across the enterprise.

The digital thread enables synchronized execution of customer orders across design, manufacturing, supply chain, and service. It also provides traceability and accountability—capabilities that are particularly critical in automotive and industrial product segments.

However, the challenge in implementing digital thread is a manufacturing reality – product data rarely sits in one place. Companies leverage primarily PLM for managing product information and leverage MES, ERP, LIMS for executing shop floor operations.  Supply chain works across planning, procurement, logistics, inventory, supplier collaboration, and risk systems leveraging SAP and Supply chain solutions. Service adds its own layer of warranty, maintenance, asset performance, and customer usage data.

For years, enterprises tried to deal with this by integrating systems. The next challenge is deeper: integrating meaning.

AI has made that gap impossible to ignore. A senior engineer can often compensate for fragmented context because she understands the product, supplier history, plant constraints, and organizational workarounds. AI agents cannot depend on tribal memory. They need structured context, governed relationships, clean taxonomies, and boundaries for action.

Manufacturers cannot scale AI in engineering and operations while leaving product context trapped in silos.

The right approach to this integration can have significant impact. For instance, the integration of ERP (SAP), PLM (Enovia), production planning and scheduling (Ortems), and MES (Apriso) enabled the creation of a digital thread for an automotive manufacturer. These systems exchanged information seamlessly, providing end-to-end visibility, component- and process-level traceability, and enabling optimization at the enterprise level rather than within individual functions.

From Digital Engineering to Enterprise Complexity Management

Today, a product is no longer a static object progressing through a linear lifecycle. It is a dynamic system of hardware, software, data, suppliers, services, and regulatory obligations. An elevator illustrates this well. It has evolved from a standalone mechanical system into a connected, integrated solution within a building management ecosystem. Continuously updated and enhanced through new features and software upgrades.

Product complexity now begins well before production. It is created through configuration options, embedded software, supplier dependencies, sustainability requirements, regional regulations, material constraints, service models, and customer-specific variants. Once that complexity enters the lifecycle, it travels. A poorly governed change in design may become a manufacturing issue, a service issue, a quality issue, or a compliance issue months later.

AI is helping engineers explore alternatives, extract requirements, accelerate validation, generate design options, and reuse knowledge. Yet these capabilities become enterprise-grade only when they are connected to a governed product context. Without that context, AI speeds up isolated tasks while leaving the organization’s complexity problem untouched.

The more advanced manufacturers use model-based approaches, digital twins, knowledge graphs, and AI-enabled engineering workbenches to detect these dependencies earlier. They simulate how design choices affect manufacturability, cost, quality, and serviceability before those choices become expensive. They treat engineering knowledge as an enterprise asset rather than something rediscovered project by project. 

The most serious manufacturing leaders are asking a harder question now: if AI agents are going to participate in engineering and operations, what enterprise context will they reason from?

Agentic Operations: The Move from Visibility to Coordinated Action

That question leads directly from digital thread to agentic operations.

Agentic AI systems can perceive context, reason through options, plan steps, trigger actions, and coordinate workflows within defined guardrails. In manufacturing, this has significant potential as the operating environment is full of decisions that are too dynamic for static dashboards and too coordination-heavy for manual workflows.

Consider a production supply chain disruption. A conventional dashboard shows a material shortage in the production plan, leading to a potential line stoppage. On the other hand, we’ve developed an Early Warning & Autonomous Control System – an intelligent supply chain management solution – that goes beyond this basic reporting. It assesses the impact of the shortage on production and customer delivery commitments and dynamically re-prioritizes the production plan to maximize capacity utilization and efficiency, with the objective of minimizing financial impact while still meeting delivery commitments.

With a human in the loop, the system then executes the next-best actions. The real business value lies in compressing the time between signal, interpretation, decision, and action.

The same logic applies to maintenance, quality, manufacturing, planning, inventory optimization, energy management, warranty analysis, and field service. These are areas where manufacturers often have enough data to see problems; yet lack the connected process intelligence to act fast enough.

The temptation will be to treat agentic AI as another layer of automation. In manufacturing, autonomy without context is dangerous. An agent making recommendations on production sequencing, quality containment, maintenance prioritization, or supplier response must operate inside a governed architecture. It needs access to the right data, visibility into constraints, clearly defined authority, audit trails, escalation paths, and human oversight proportional to business risk.

The operating question is: which decisions can be safely accelerated because the enterprise has the data, process discipline, governance, and human accountability to support them?

Fig 2: From Digital Thread to Agentic Operations

The Workforce Is the Real Adoption Layer

Technology programs often underestimate the human system and, in our experience, this is one of the biggest challenges of Digital Transformation in the manufacturing organization. A supervisor who does not trust the data will work around the system. An engineer who sees PLM workflows as administrative overhead will keep using spreadsheets. A planner who cannot understand an AI recommendation will ignore it under pressure. A plant manager who sees digital transformation as corporate theatre will protect daily output first.

This is why manufacturing transformation depends on operating routines, incentives, skills, and trust. As AI systems take on more coordination and execution tasks, organizations will need to redesign roles around supervision, exception management, decision accountability, and continuous improvement. The workforce will need to become comfortable working with intelligent systems, questioning them, guiding them, and improving them over time. Increasingly, change management is an integral part of the digital transformation programs. 

The CXO Agenda: Build the Operating Architecture for Intelligent Manufacturing

For manufacturing CXOs, the priority is not to choose between PLM modernization, smart manufacturing, planning tools, digital twins, GenAI, and agentic AI. These are connected parts of one operating architecture.

The digital thread provides trusted lifecycle context. PLM governs product truth. Smart manufacturing connects operational reality. Digital twins simulate decisions. GenAI changes how knowledge is accessed and created. Agentic AI coordinates action across workflows.

The sequencing will differ by enterprise. One manufacturer may begin with engineering change velocity. Another may focus on supplier-driven quality variation. A third may tackle software-defined product complexity. Another may target downtime, service performance, or regulatory evidence. The architecture can be common, but the value thesis must be specific. 

The uncomfortable truth is that AI will expose fragmented enterprises faster than it fixes them. If product data is inconsistent, AI will surface inconsistency faster. If processes are poorly governed, agents will accelerate poor decisions. If engineering and manufacturing operate with different versions of truth, GenAI will produce confident confusion. If workforce adoption is treated as a training checklist, expensive systems will sit on top of unchanged behavior.

The manufacturing enterprise is becoming too complex to run through disconnected systems and retrospective intelligence. Leaders now need a digital engineering foundation that can understand the product, sense the operation, govern the decision, and coordinate the response.

That is the real journey from digital thread to agentic operations: building a manufacturing enterprise that can reason, adapt, and execute with discipline.

Author:

Nitin Kumar Kalothia
Associate Partner – Business Consulting Group
ITC Infotech