Enterprise-wide AI transformation starts with change management

Technology leaders are investing heavily in AI, yet many initiatives struggle to move beyond pilots, with Gartner finding only 28% of AI use cases in infrastructure and operations fully succeed and meet ROI expectations, while 20% fail outright. A CIO and CDO argues that success depends on operational readiness, governance, employee adoption, and change management, not just technical capabilities.

Technology leaders are facing a sobering reality: They’re investing heavily in AI, yet many initiatives continue to struggle to move beyond experimentation and pilot programs. For example, Gartner found only 28% https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-says-artificial-intelligence-projects-in-infrastructure-and-operations-stall-ahead-of-meaningful-roi-returns of AI use cases in infrastructure and operations fully succeed and meet ROI expectations, while 20% fail outright. The conversation around AI often focuses on models, tools and technical capabilities. Those decisions matter, but in my experience, they are rarely the only factors that determine success. The organizations realizing meaningful value from AI are also focused on operational readiness, governance, employee adoption and measurable outcomes. As both CIO and CDO, I spend a lot of time helping our organization navigate AI adoption while balancing the needs of our internal teams, our clients and running 24×7 secure operations. What I have learned is that AI transformation depends on how well the organization understands its data, improves its business processes and prepares people to work differently. I sometimes describe my role as being the organization’s traffic light. The green lights are easy – these are moments when the right answer is to accelerate. There are also moments when we need to slow down. As leaders, we must assess when we need to focus on the fundamentals and make sure the organization is ready for what comes next. And the most important decisions are the red lights – when we prevent the organization from spending time, money and energy on the wrong things. One common misconception about AI transformation is that deployment automatically creates adoption. In practice, adoption happens when employees understand how the technology improves their work and have confidence in how it fits into their day-to-day responsibilities. I have seen AI pilots work well with small groups of users and then encounter challenges when expanded across larger teams. The technology may perform as expected, but the operating environment changes. Teams follow different workflows. Information is managed differently across functions. Employees have different levels of trust in the data. Success is not always measured the same way. These are readiness, process and change management issues. We saw similar lessons during our own transformation work. As part of a broader modernization program, we consolidated more than 50 engineering tools into one software delivery platform supporting thousands of developers. The technical migration mattered, but the bigger effort was helping teams adopt new ways of working and establish common practices. Anyone who has asked developers to move away from their favorite tools knows that change management is real. That experience reinforced a lesson: Transformation succeeds when people understand the value of the change, have the right support and can see how it improves the work they do every day. The same principle applies to AI. When we began introducing AI capabilities internally, we avoided a broad rollout from day one. Rolling AI out to thousands of employees is a process of education, adoption support and continuous learning. We introduced capabilities in phases, helped employees understand use cases relevant to their role and gave teams room to build confidence over time. Different teams adopt AI differently, so we found that cohort-based deployment and tailored change management created better long-term adoption than broad enterprise-wide rollouts. Pilots often succeed because the variables are limited. Production environments introduce the realities of the enterprise: inconsistent processes, disconnected data, unclear ownership and varying levels of employee readiness. In many cases, issues that surface during scaling can be traced back to operating model decisions, process gaps or unclear expectations. Employees need to understand where AI fits, when human judgment remains essential and how success will be measured. Without that clarity, scaling becomes much harder. The most successful AI transformations start before AI is introduced. They begin with understanding where employees experience friction. In most enterprises, those opportunities are not difficult to find. Repetitive administrative work and manual handoffs consume time and slow the business down. Employees directly in the workflows have the clearest view of where these issues exist. When we launched our own efficiency and transformation program, we deliberately did not start with AI. We started by evaluating our data, reviewing business processes and identifying opportunities to simplify how work was performed. We found that simplifying and standardizing workflows before introducing AI significantly reduced complexity during deployment. Rather than asking AI to compensate for fragmented processes, we focused first on creating a consistent operational foundation. We focused first on process improvement, automation and operational discipline. Once those foundations were in place, we began layering AI into the environment. AI outcomes are heavily influenced by the quality of the processes and the data along with the governance structures supporting them. If the underlying process is inconsistent, AI will struggle to create consistent value. If the process is understood, governed and measurable, AI has a much stronger foundation. I often say that good data and good processes deliver good AI outcomes. That continues to hold true regardless of the model or technology being deployed. The real challenge is making sure employees know what AI is using, where it fits in the workflow and when they should rely on the output. If that is unclear, adoption slows. People may not trust the answer, may use the tool inconsistently or may avoid changing how work gets done. Before scaling AI, leaders need to answer a few basic questions. What problem are we solving? Is the process consistent enough? Is the data reliable enough? Where does human judgment still matter? And how will we know whether the tool is improving the work? Those questions determine whether AI becomes part of how teams operate. AI programs often lose momentum when leaders measure activity instead of impact. 63% of high-maturity organizations https://www.gartner.com/en/newsroom/press-releases/2025-06-30-gartner-survey-finds-forty-five-percent-of-organizations-with-high-artificial-intelligence-maturity-keep-artificial-intelligence-projects-operational-for-at-least-three-years implement formal metrics to evaluate transformation efforts. Leaders often track how many employees have access to AI, how many licenses have been provisioned or how many use cases have been launched. Those metrics can be useful, but they do not always show whether the organization is creating business value. Activity is not the same as impact. The more meaningful indicators are instead tied to operational performance: support ticket volumes, incident reduction, productivity improvements, user experience, cycle times and service quality. We have seen the value of this approach firsthand. As part of our transformation program, we standardized service delivery processes and moved hundreds of teams onto a common service management platform. In our own experience, process improvements and platform consolidation initially reduced support ticket volumes by approximately 30%. After that foundation was established, additional automation and AI capabilities helped drive reductions closer to 70%. The initial improvement came from better processes and greater operational consistency. Automation and AI then helped accelerate the results. That is the pattern leaders should look for: Identify where work slows down, improve the process, establish accountability and introduce AI where the environment is ready to support it. This approach also helps build trust. Employees can see the value being created. Leaders can measure progress. Teams can learn from early deployments before scaling more broadly. Technology adoption has always been closely connected to people. Employees are more likely to embrace change when they understand how technology helps them be more effective. They need practical experience, clear expectations and opportunities to learn. AI introduces new ways of working, and organizations need to prepare employees for that shift. In our own organization, we encouraged every employee to establish an AI-related learning goal because familiarity with emerging technologies is becoming part of every role. Some goals were simple. Some were more advanced. The important point was creating a culture where people continue to learn and understand how AI applies to their work versus forcing AI activity broadly all at once. As AI becomes more embedded in enterprise operations, organizations with strong foundations in governance, process discipline and workforce readiness will be better positioned to capture long-term value. The companies realizing the greatest value from AI are investing in technology while also strengthening the operating models, information management practices and employee capabilities that support adoption. Sustainable transformation requires attention to people, processes, data and technology. Preparing people, building trust and creating clear operating models remain central to any successful AI strategy. This article is published as part of the Foundry Expert Contributor Network. Want to join?