From dashboards to decisions
Dayan Rodriguez, Corporate Vice President of Manufacturing & Mobility at Microsoft, learned one of the most important lessons of industrial transformation long before he joined the technology company. Early in his career, working alongside operators and plant managers on manufacturing shop floors, he discovered that production teams have little interest in how sophisticated a technology is. They care whether it helps them solve today’s problem faster, more safely and more efficiently than yesterday.
That experience continues to shape Rodriguez’s view of industrial AI. He believes the manufacturers pulling ahead have stopped asking whether the technology works and started asking a more important question: how do you embed intelligence into everyday operations, earn the trust of the workforce and scale successful deployments across an entire organization?
“The manufacturers scaling AI start by understanding how work happens on the factory floor, across the supply chain and within procurement, then embed intelligence directly into those workflows,” Rodriguez says. “Not as an add-on, but as part of how decisions get made. A use case that lives in a dashboard nobody looks at isn’t a success. It’s a cost.”
That shift reflects how rapidly the manufacturing conversation has evolved. Where organizations once focused on AI’s potential, they are increasingly looking for measurable operational outcomes that can be replicated across multiple sites. “It really has changed,” Rodriguez explains. “A year ago, people wanted to talk about what AI could do. Today they want to know what it’s doing and what it’s costing or saving them.
“I’m no longer being asked where to start. I’m being asked how to move faster and how to make sure things scale. Even who’s in the room has changed. AI conversations are no longer led solely by innovation or IT teams. I’m regularly talking with plant operations leaders, supply chain directors and CFOs. Those conversations don’t tolerate vague technology promises. They want to know what impact looks like and when.”
Rodriguez believes that shift is reflected across the wider market. He points to Deloitte’s 2026 State of AI in the Enterprise report, which found that almost three-quarters of enterprise organizations expect to deploy agentic AI within the next two years, while physical AI adoption is forecast to reach 80 per cent, with manufacturing expected to lead much of that transition. For him, those figures signal that manufacturers are moving beyond experimentation and beginning to treat AI as part of their operating model rather than another digital initiative.
That changing mindset is already evident in practical deployments. Rodriguez highlights TK Elevator, whose AI agents automatically assemble equipment history, service records and safety guidance before every technician visit. Instead of searching multiple systems for information, technicians arrive equipped with the context they need to make faster, better-informed decisions. It is a simple example of a wider principle that runs throughout manufacturing today: the greatest value of AI often comes not from replacing human judgement, but from ensuring the right information reaches the right person at exactly the right time.
Scaling is where the real challenge begins
If manufacturers have largely answered the question of whether AI can deliver operational value, the next challenge is determining whether that value can be reproduced consistently across an entire business. That is proving considerably more difficult. The barriers rarely lie in the AI models themselves. Instead, they emerge as organizations attempt to apply successful projects across factories that have evolved independently, each with its own equipment, processes and ways of working.
“I hear this one constantly,” Rodriguez says. “Scaling AI is harder than building it, and most organizations learn that the hard way. A team gets a great result at one plant. The AI is working, operators trust it and the numbers look good. Leadership says, ‘Let’s roll it out everywhere.’ Then the problems surface. Plant two has different equipment, different data standards and different local processes. What worked at plant one doesn’t transfer cleanly, so instead of scaling, you’re rebuilding over and over again.”
Those differences are familiar to every manufacturing organization. Production assets have often been added over decades, creating environments where legacy machinery sits alongside modern automation, each producing data differently and supporting different operational practices. As AI becomes more deeply embedded within production, however, technical complexity is only part of the equation. Rodriguez believes the organizations making the greatest progress are treating AI as an enterprise transformation rather than a series of technology deployments.
“The technical fragmentation is real, but the organizational challenge is just as significant,” he explains. “Scaling AI requires alignment across plant operations, IT, engineering and business leadership. When those functions aren’t pulling together around a shared outcome, AI ends up fragmented, with each site doing its own thing and no ability to compound value across the enterprise.”
That shift from isolated projects to enterprise capability is increasingly influencing how manufacturers evaluate AI investments. Rather than asking whether a single pilot delivered measurable results, organizations are beginning to ask whether the deployment has established a repeatable approach that can accelerate future implementations. In that respect, common data standards, governance and operating models become strategic assets because they allow knowledge and capability to move with the technology.
Rodriguez points to Krones as an example. Working with Microsoft and partners including NVIDIA and Ansys, the company integrated AI-driven fluid simulation into a digital twin of its filling line, reducing simulation times from four hours to under five minutes, a 95 percent improvement. Beyond the engineering gains, the project demonstrated how solutions built on a common platform can be replicated across global engineering operations rather than remaining confined to a single site.
“Prove it, build it to scale, then expand,” Rodriguez says. “That’s the model.” It is a simple philosophy, but one that captures the direction industrial AI is now taking. The manufacturers pulling ahead are no longer judged by the number of pilots they launch, but by their ability to turn individual successes into repeatable capabilities that strengthen the business with every deployment. Scaling, ultimately, has become an organizational capability rather than a purely technical exercise.
Trust becomes the competitive advantage
If operationalizing AI depends on scaling it successfully, sustaining that progress depends on trust. Manufacturers can invest in increasingly sophisticated models and automation technologies, but those investments only create value when employees are confident enough to act on the recommendations AI provides. According to Rodriguez, that confidence is earned long before a model is deployed. It begins with reliable data, grows through practical experience and is reinforced by governance that gives people confidence in how intelligent systems are operating.
“AI can only be as good as the data behind it,” Rodriguez says. “If your data is incomplete, disconnected or inconsistent with what actually happens on the factory floor, your AI will generate insights that aren’t reliable. The moment an operator acts on a bad recommendation, and something goes wrong, you’ve lost their trust. That’s very hard to get back.”
Rather than waiting until every legacy system has been modernized, Rodriguez advocates building that confidence incrementally. Manufacturers should focus first on creating trusted information around high-value operational use cases before extending AI across the wider organization. Microsoft Fabric IQ supports that approach by bringing together operational, engineering and enterprise data to provide AI systems with the industrial context needed to generate meaningful recommendations. Rodriguez points to manufacturers including BMW and Denso, where establishing a unified industrial data foundation became a critical element of broader cloud and AI transformation programs.
Trust, however, extends beyond data. Manufacturing is entering a period in which decades of operational experience are leaving the workforce through retirement, creating an urgent need to preserve institutional knowledge before it disappears. Rodriguez believes this represents one of AI’s greatest opportunities, not replacing experienced engineers and technicians, but making their expertise available to colleagues across the organization.
“I have deep respect for what experienced operators and technicians actually know,” he explains. “The pattern recognition built from twenty years on a specific production line is genuinely irreplaceable. The problem is that knowledge retires when they do. One of the most important things AI can do is capture and distribute that expertise before it walks out the door.”
Microsoft Foundry IQ reflects that philosophy by combining operational history, procedures and engineering knowledge into AI-driven workflows that support less experienced employees. Rodriguez highlights TK Elevator, where AI agents prepare technicians before each service visit by assembling equipment history, maintenance records and safety guidance, while also capturing the knowledge generated during every visit so it becomes available to the wider service enterprise. Rather than replacing human expertise, AI enables organizations to retain, share and continuously expand it.
As intelligent systems become embedded within day-to-day manufacturing operations, governance is evolving from a compliance requirement into an operational capability. Rodriguez believes the most successful enterprises no longer view governance as the responsibility of IT alone. Instead, they are establishing cross-functional frameworks that define how AI is deployed, how decisions are monitored and where accountability ultimately resides.
“The most mature manufacturers treat governance as a competitive capability,” he says. “When you demonstrate to customers, regulators and your own workforce that your AI operates responsibly, within defined boundaries and with full auditability, that becomes real differentiation. Trust is the new competitive advantage in industrial AI.”
That conclusion is reinforced by Microsoft’s 2026 Work Trend Index, which found that organizational factors such as leadership, culture and workforce development account for 67 percent of AI impact, compared with 32 percent driven by individual behavior alone. The finding neatly reinforces Rodriguez’s central argument. The manufacturers defining the next phase of industrial transformation will not necessarily be those deploying the most AI. They will be the enterprises that create the confidence, capability and organizational discipline to make intelligence part of everyday manufacturing.
Redesigning how work gets done
Looking ahead, Rodriguez expects AI to influence every part of the manufacturing value chain, from maintenance and quality through to engineering, production planning and supply chain management. He believes the greatest competitive advantage, however, will not come from deploying the largest number of AI applications, but from connecting intelligence across the enterprise so information, expertise and decision making become part of a single operating model.
He points to developments such as agentic AI, digital twins and physical AI as evidence that manufacturing is moving beyond systems that simply predict events towards technologies capable of recommending and, in carefully defined situations, executing actions. Even then, he argues, successful companies will continue to keep people firmly at the center of operational decision making, using AI to augment judgement rather than replace it.
“The manufacturers building sustainable AI strategies have stopped thinking about AI as a tool and started thinking about it as part of how their business operates,” Rodriguez concludes. “The enterprises stuck in disconnected use cases are still asking where the next AI project is. The frontier manufacturers are asking how they redesign how work gets done. Those are very different questions, and they lead to very different outcomes.”
Rodriguez’s message is ultimately less about artificial intelligence than organizational transformation. Throughout the conversation he returns to the operational disciplines that determine whether technology creates lasting value: trusted data, aligned organizations, scalable operating models and empowered people. Manufacturers have largely demonstrated that AI can work. The next generation of industrial leaders will be distinguished by something more difficult, embedding intelligence so deeply into everyday operations that it becomes simply another way manufacturing gets done.
