AI transformation is the CEO’s job

AI transformation is fundamentally a people challenge that must be led by the CEO, not a technology project, according to Star, which advocates for 'organizational recomposition'—reassembling existing talent through upskilling and lateral moves rather than simply shrinking the company. The piece warns that culture must be used as a vehicle for change, not replaced, and that leaders must manage the pace of adoption carefully, citing Meta CEO Mark Zuckerberg's admission of mistakes in restructuring too quickly and a client example where an engineer ran up $70,000 in AI token costs without controls.

Most conversations about AI https://www.fastcompany.com/section/artificial-intelligence -ready leadership focus on the technology— which workflows to rebuild, which tools to buy. True, the CEO must endorse these decisions but that isn’t their role. In the AI era, their role is one of people transformation and the readiness that matters most is their own willingness to lead it first. At Star, we call this “organizational recomposition,” and it is by far a harder job than many realize. It means taking the talent, skills, and capabilities of the people you have today, decomposing them, and reassembling the organization into something that isn’t much bigger but is different. This is achieved by upskilling, laterally moving people into work that didn’t exist a year ago, or simply through new ways of working that AI has only just made possible. Recomposition does not mean a smaller company. It means a more capable one, built substantially with the people you already have on your roster. However, in all honesty, organizational recomposition is also about understanding how to remove those employees who are not ready to get on board with AI transformation. This isn’t a technology project. It’s a people’s decision, and it sits with the CEO. Change is never easy and there are two factors—the culture of the business and the pace of change—that must be handled correctly for transformation to be a genuine success. Employees are often seen as being resistant to change, but much of this resistance isn’t about tool adoption; it’s about not wanting the very culture of the business to change. After all, the key cultural pillars on which any organization is based are essential to its core brand. It’s what customers, stakeholders, and employees perceive the business as and judge you on. To lose that core pillar would fundamentally change the organization. Instead, use the culture as a vehicle for change. Work from the existing cultural pillars so that the transformation feels like a natural evolution of what the company already is, not a replacement of it. After all, if recomposition makes the company less itself, you haven’t transformed it, you’ve replaced it, and your people and customers will know. The pace of change is where most leaders still struggle because there is currently no proven method as to which AI tools to adopt, and just as importantly, which to drop. AI is not a unique point in time; most CEOs will recognize that we have lived through similar transformations before. For example, when the cloud arrived the companies that won were not the ones that moved their servers into data centers the fastest. Instead, the real winners were those whose leaders treated the cloud for what it was: a change in how, when, and where people worked. It’s worth remembering that cloud adoption also taught us a lesson around cost. The cloud promised radically more efficient use of capital and resources, and it delivered, right up until success arrived. More users, more usage, and the running costs curved past what you’d have paid to host the thing yourself. It is easy to see that AI is on the same curve, only faster, and at a far greater scale than anyone imagined. Mark Zuckerberg admitted recently that https://www.reuters.com/business/metas-zuckerberg-admits-mistakes-made-ai-transformation-2026-06-12/ Meta made “mistakes” https://www.reuters.com/business/metas-zuckerberg-admits-mistakes-made-ai-transformation-2026-06-12/ restructuring around AI faster than it could absorb. It’s an issue we are increasingly seeing with our own clients: many worry about being too slow, while almost none worry about being too fast. One client recently told me how they had instructed engineers to start using AI and make it a daily habit. That was until they discovered one had run up $70,000 in tokens. Usage-based AI costs can escalate quickly when teams are encouraged to adopt AI without clear controls, monitoring, or guardrails. These costs then become a barrier to change. Pacing the rate of adoption is one of the calls that only the CEO can make. Ultimately, a CEO can delegate the infrastructure and the technology to others, but what can’t be delegated are the tough decisions around culture and trajectory. That is because in turbulent times people look to the CEO to be the one who can answer calmly and reassuringly when no one else can. And that takes time. It’s not a sprint; we’re in an ultra-marathon situation. When it comes to AI transformation there’s not going to be a quick win or an easy answer. The companies that come out the other side won’t necessarily be bigger, in fact they may well be the same size, but they will be a great deal more capable. And that is the win worth recomposing for. Michael Schreibmann is CEO and cofounder of Star.