The Great ITSM Reset: Why AI Is Forcing IT to Go Back to Basics

Enthusiasm for artificial intelligence in IT has long been out of control. Business leaders, executives, and operational teams are pushing to integrate AI into every service management process, convinced it will deliver immediate efficiency, automation, and cost savings. The reality, once AI applications are deployed, is often bleakly different: pilots that cannot be scaled, ROI that is difficult to demonstrate, AI agents that fail where processes are fragile.

In most cases, the root cause of these difficulties lies not in the technology itself, but in the organizational context in which it is placed.

The actual effect of AI on ITSM is the opposite of today’s prevailing narrative: rather than replacing it, AI is bringing it back to fundamentals. Documented workflows, clear ownership, adherence to ITIL best practices, and operational maturity — without these foundations, any AI readiness initiative in ITSM is destined to generate costs and complexity instead of value.

1. What Does AI Readiness in ITSM Mean?

ITSM is a system of record — it manages structured workflows and contains the organization’s operational context. For all these reasons, it should naturally serve as the foundation for AI in IT operations. However, for this potential to translate into concrete value, ITSM must exhibit certain characteristics. AI readiness in ITSM means having digitally documented processes, clear ownership, adherence to ITIL best practices, and measurable indicators of operational maturity. These are the minimum conditions for an AI initiative in ITSM to generate value rather than additional costs.

The Current State of ITSM: Fragile Implementations Under AI Pressure

A growing number of organizations are now extending their ITSM platform to workflows outside traditional IT: HR, facilities, security. In this sense, ITSM constitutes the backbone of business operations — the spine of the enterprise.

The problem is that, in many organizations, this backbone is fragile. Four recurring issues stand out:

• Undocumented processes: the same workflows vary from team to team, and operating procedures seem to exist only in the heads of the most experienced people.
• Undefined ownership: it is unclear who is responsible for a service, a process, or the quality of a dataset.
• Poor adherence to ITIL best practices: incident, problem, change, and service request management do not follow consistent, coherent patterns over time.
• Uneven operational maturity: some teams work with structured processes and formalized procedures, while others operate reactively and without standardization. The result is an organization that, on the whole, is pulled down to function at the level of its least mature team.

These are pre-existing problems that have, until now, remained manageable: human operators, drawing on experience and common sense, were able to bridge process ambiguities and adapt to exceptions. AI changes the rules: an automated agent cannot make intelligent decisions about an unwritten process, cannot fill ownership gaps, and cannot compensate for an inadequate level of organizational maturity.

In the absence of these preconditions — documented workflows, clear ownership, adherence to ITIL best practices, operational maturity — AI exposes process instability rather than compensating for it. What was previously a weakness that could be more or less easily absorbed into the process now becomes an operational blocker.

The Promise of AI and the Reality of the Numbers

Market data from 2025 tells a clear story of contradiction. On one hand, AI adoption is widespread: according to recent research by EasyVista, 95% of enterprise organizations already use AI in ITSM in some form, and the figure remains very high — at 90% — even among SMBs. On the other hand, the financial results tell a very different story.

A Gartner survey of more than 500 CIOs found that 72% of organizations are barely breaking even or actually losing money on their AI investments. BCG’s Build for the Future 2025 study of 1,250 companies is even more stark: only 5% create substantial value by applying AI at scale, while 60% experience no material value generation whatsoever. Research from the MIT NANDA Initiative, cited by Fortune, further indicates that 95% of enterprise GenAI pilot projects generate no measurable return on the income statement.

The real question to ask, therefore, is not so much about the functionality of AI itself, but about the reasons why, once applied, it so often fails to produce concrete results. The answer is uncomfortable and must be addressed directly and promptly: AI is failing for operational and process-related reasons, not merely technological ones.

AI Is Not a Shortcut — It Is an Amplifier

One of the most widespread and most mistaken beliefs is that AI can compensate for budget shortfalls, lack of skills, or insufficient rigor in process execution. This logic is flawed: AI is genuinely useful when it accelerates work that teams already know how to do well — not when it attempts to replace skills that do not exist. Applied to mature processes, it amplifies quality. Applied to fragile processes, it amplifies disorder.

Keith Andes, Head of Product Marketing at EasyVista, captures the situation in a striking image: “If incident routing is inconsistent today, AI will simply route incorrectly even faster. If the Configuration Management Database (CMDB) — the central repository that tracks a company’s IT assets and their relationships — is messy, the insights generated by AI will simply reflect the background noise.” The old principle of garbage in, garbage out applies, with an added complication: AI output is wrapped in an air of authority that makes the problem harder to detect.

Boston Consulting Group’s 10–20–70 principle quantifies this dynamic: AI success depends 10% on algorithms, 20% on data and technology, and 70% on people, processes, and culture. McKinsey’s State of AI 2025 confirms the same logic: companies that redesign end-to-end workflows before selecting models or tools are nearly three times more likely to achieve significant financial returns than those that simply layer AI on top of existing processes.

The Four Pillars of AI Readiness in ITSM

The Great ITSM Reset represents a deliberate and targeted return to ITSM fundamentals — a necessary condition for AI to actually function. It is a consolidation process that involves people, processes, and data, and that requires time and discipline. But it is also the only path that allows AI investments to be transformed into measurable, sustainable results over time. The four pillars on which this work rests are deeply interconnected.

1. Digitally Documented Workflows

An AI agent cannot execute a task it cannot see. For every critical service, the steps, decisions, triggers, and closure criteria must be represented within the ITSM platform in structured form. This is the concept of the digital twin of work: a representation of how work actually happens, not how one would like it to happen.

2. Clear Ownership of Services, Processes, and Data

Without a clearly identified owner, every service is destined to degrade over time. AI only makes the problem worse: without clear ownership, no one is responsible for validating outputs, intervening when suggestions are systematically wrong, or authorizing the application of AI to new processes and areas of work. Defining ownership — by service, process, and knowledge base area — is a governance prerequisite before it is a technological one.

3. Adherence to ITIL Best Practices

ITIL practices are the common language that makes processes predictable, measurable, and repeatable. When incident, problem, change, and configuration management follow consistent patterns, the ITSM platform generates high-quality structured data — and it is precisely this data that AI uses as its raw material. The ITIL 4 framework itself emphasizes that value does not emerge from the formal adoption of processes, but from their consistency over time.

4. Measurable Operational Maturity

Maturity is not a status that, once declared, can be taken for granted; it is a measurable condition driven by consistent resolution times, controlled reopening rates, CMDB accuracy, and verified compliance. A recent Gartner survey shows a clear difference between mature and non-mature organizations: in the former, 45% of AI initiatives remain in production for at least three years, compared to 20% in low-maturity organizations.

How to Begin the Reset, in Practical Terms

Carrying out a reset means making the introduction and use of AI sustainable without slowing its adoption. The goal is to create the operational conditions for AI to generate real, measurable value — starting where it is easiest to demonstrate results and proceeding through successive expansions. Three practical steps help set the course:

1. Identify one or two high-volume, already stable workflows. These are the natural candidates for the first AI use cases: ticket routing on reliable categories, automatic knowledge summarization from already verified content, and automatic ticket summarization based on structured data.
2. Clean up first, then automate, then apply AI last. The first step is to digitize the workflow in the system of record. The second is to automate everything that can be managed with simple, predictable “if X happens, do Y” instructions. Only for activities too complex to be resolved with rules of this kind does it make sense to introduce AI.
3. Establish value metrics. Without an initial set of KPIs — mean time to resolution, self-service deflection rate, accuracy, or user satisfaction — there is no way to know whether AI is producing results.

Technologies such as EasyVista’s EV Pulse AI are designed to support and enhance precisely this kind of approach, integrating AI within structured ITSM workflows through defined use cases, clearly readable outcome metrics, and constant human oversight of every critical decision.

AI does not make process discipline superfluous — on the contrary, it requires it as a condition of operation. From this perspective, the Great ITSM Reset is itself an element of innovation: the path through which IT organizations can develop their AI readiness in ITSM and evolve in the years to come.

FAQs

1. What does AI readiness in ITSM mean? 

It means having digitally documented processes, clear ownership, adherence to ITIL best practices, and measurable indicators of operational maturity. These are the minimum conditions for an AI initiative in ITSM to generate value rather than additional costs.

2. Why do so many AI initiatives in ITSM fail? 

Because AI is applied on top of fragile processes, inconsistent data, and undocumented workflows. Studies by BCG and McKinsey indicate that in 60% of companies AI generates no material value, and 95% of GenAI pilots produce no measurable ROI. The cause is almost never technological — it is almost always organizational.

3. Do processes need to be perfect before starting with AI? 

No, but it is necessary to start from areas where processes are already stable. AI works as an accelerator for consistent, robust workflows. It does not work as a solution to chaotic workflows. The principle is: clean up first, then automate with rules, then apply AI only where it is truly needed.

4. What is the relationship between ITIL and AI? 

ITIL practices provide the common language and data structure necessary for AI to operate reliably. The higher the ITIL maturity, the more solid the foundation on which to build intelligent automations and AI agents.