The Framework That Was Always Ready for AI
There is a conversation happening in boardrooms and technology forums right now that sounds new but is not.
Executives are asking how to govern AI at enterprise scale. How to deploy it responsibly, measure its impact, prevent it from creating more chaos than value. How to move from a collection of pilots to a transformation that actually sticks.
ITIL has been answering that question for thirty years. ITIL (Version 5) is the first iteration that answers it explicitly, and in manufacturing and operational technology environments specifically, the timing could not be more important.
Why Version 5 Matters, and Why Now
ITIL (Version 5) was not released because the framework was broken. It was released because the world it operates in changed faster than any previous update could keep pace with.
The cloud turned infrastructure into an on-demand service. AI began making decisions that used to require humans. Digital product and service management stopped being a centralized IT function and became something that happened, whether IT was involved or not, across every business unit in the enterprise. In manufacturing environments, this happened at the plant floor level, where operational technology teams were adopting intelligent systems without the governance architecture to manage them safely.
ITIL (Version 5) acknowledges this directly. It restores the profession to its original, fuller scope: not just managing IT services, but managing technology-enabled products and services through their entire lifecycle, wherever in the organization they exist. In Industry 5.0 terms, that means AI systems, digital twins, IIoT platforms, and analytics capabilities that influence physical outcomes and human decisions in real time.
For those of us who have been making this argument in practice for years, that ITIL was always a value creation framework, not just a process discipline, Version 5 is the official confirmation.
The AI Governance Problem ITIL (Version 5) Was Built For
Let me describe the challenge precisely, because it is more specific than "AI is complicated."
In manufacturing environments, the governance gap sits at the intersection of IT, OT, and the business. OT standards govern how machines behave. What they do not govern is who owns an AI-driven recommendation after it leaves the model. How risk is accepted across functional boundaries. How an intelligent system evolves after go-live when the production environment it was trained on has changed. How a human operator at two in the morning knows when to trust the output and when to override it.
I’ll give an example of what can happen when that question has no clear answer. A quality inspection model at a regulated manufacturing facility was deployed with strong initial accuracy. Over fourteen months, the product formulation shifted incrementally within specification. Nobody retrained the model. Nobody owned the performance review cadence. By the time the issue surfaced during a regulatory audit, the model had been flagging compliant product and passing non-compliant product at rates that could not be easily explained to the inspector. The technical team could describe what the model did. Nobody could demonstrate who was responsible for its ongoing performance. That gap cost months of remediation effort and significant regulatory credibility. It was entirely preventable with basic service governance.
These are not technology questions. They are service management questions. And ITIL (Version 5) gives us the vocabulary and structure to answer them, now with explicit guidance for managing AI as a managed capability, not a project.
The framework's treatment of the product and service lifecycle is particularly important here. AI systems are not static deployments. They are living services. They have consumers whose needs evolve. They have dependencies that shift. They require continual improvement not as a best practice but as a survival requirement. A model that is not monitored, retrained, and governed will drift. And a drifted model in a safety-critical or regulated environment is not just a technology failure. It is a governance failure with real operational and human consequences.
What "AI-Native" Actually Means in ITIL Terms
ITIL (Version 5) introduces explicit guidance for AI-native organizations, those seeking to integrate AI across their products and services rather than bolt it on as a feature. In manufacturing, this distinction is critical.
Most organizations using the phrase "AI-native" are actually AI-aspirational. They have AI tools, pilots, and ambitions. What they lack is the operational and governance infrastructure that would make AI a reliable, measurable, improvable service. Good automation needs good data, configuration, and knowledge management. That is not a new idea. It is the foundational ITIL argument applied to a new context.
Before you can automate intelligently, you need to manage what the automation depends on. In a pharmaceutical manufacturing environment, that means EBR data that is structured and accessible, not locked in scanned PDFs. In a process manufacturing environment, it means historian data that is governed and connected, not siloed by site. In any regulated environment, it means audit trails that demonstrate not just what the system decided but why, and what a human did with that recommendation.
Organizations calling themselves AI-native without building that foundation are not AI-native. They are AI-aspirational. The distinction has real financial and regulatory consequences.
The Data Foundation Argument Has Not Changed, It Has Deepened
One of the most important things ITIL (Version 5) reinforces is something that has always been true but is now impossible to ignore: data governance is service governance.
The reason so many AI transformations stall is not model quality or compute capacity. It is that the data the models depend on is ungoverned, unstructured, and untrusted. ITIL (Version 5) treats information and knowledge management not as peripheral practices but as core capabilities, prerequisites for everything else in the framework. When you apply that lens to AI in manufacturing, the investment conversation changes entirely. You are not talking about model licensing or GPU clusters. You are talking about data governance, configuration management, and the information architecture that determines whether AI can create value or whether it creates expensive, well-packaged noise.
That is a CFO-level conversation. It is also an ITIL conversation. And Version 5 makes the connection explicit in a way that previous versions left implicit.
A Word to the ITIL Community
ITIL (Version 5) was released in 2026 because the timing was right. The AI governance conversation is open. The regulatory environment, NIS2, IEC 62443, ISO 42001, FDA expectations for AI-assisted manufacturing, is forcing accountability onto executives who were previously insulated from technology risk. Organizations are searching for structure that connects infrastructure investment to business outcomes while preserving human authority over consequential decisions.
We have that structure. We have always had it. Now the framework itself says so, in language that speaks directly to the AI era and to the operational environments where governance failures have the most serious consequences.
The conversation has never been more ours to lead.
Enjoyed this post? Join the conversation by leaving a comment or sharing your thoughts below, we’d love to hear your experiences and perspectives. Don’t forget to explore our upcoming events for more opportunities to learn and connect, and visit the forum to continue the discussion.
