Why Most Companies Are Getting AI Wrong — And What McKinsey’s New Manifesto Gets Right
We are living in an era of unprecedented technological investment, yet for many enterprise leaders, the promised returns of artificial intelligence remain stubbornly out of reach. While nearly every major corporation is rushing to deploy generative and agentic AI tools, the reality on the ground tells a different story. According to recent research from KPMG, only a small fraction of organizations—about one in ten—possess the talent, governance, and operating discipline necessary to achieve compounding returns on their AI spending. The rest, as one industry analyst aptly described it, are simply “spending and hoping”.
It is against this backdrop of widespread frustration and stalled pilot programs that McKinsey & Company has released its latest briefing, The AI Transformation Manifesto. Excerpted from the second edition of their book Rewired, the manifesto distills the consulting firm’s observations from hundreds of large-scale tech and AI transformations into twelve core themes. Rather than another breathless celebration of technological capability, the document serves as a sobering checklist for change. It forces executives to confront an uncomfortable truth: AI success is fundamentally an organizational and leadership challenge, not a technological one.
The Illusion of Technology as Advantage
Perhaps the most striking assertion in McKinsey’s manifesto is its first: technology alone does not create a competitive advantage. The tools themselves—large language models, agentic frameworks, and advanced data platforms—are broadly available to anyone with a budget. Instead, the companies that are truly winning with AI are those that have built enduring organizational capabilities to harness any new technology effectively.
This aligns perfectly with broader industry data. A recent survey of corporate executives revealed that half of all leaders cite organizational and cultural hurdles as the primary barrier to AI success, compared to only 40 percent who point to technical limitations. The problem is rarely the software; it is the system into which the software is being deployed.
McKinsey highlights that successful organizations are abandoning the scattershot approach of pursuing long lists of isolated AI use cases. Instead, they are applying what the manifesto calls the “economic leverage” rule. These companies focus their efforts on one to three core business domains—such as process yield in the mining sector or supply chain integration in the automotive industry—where AI can deliver a massive impact, averaging a 20 percent EBITDA uplift. They understand that if the value being created does not fundamentally move the business forward, the implementation strategy is flawed.
The Leadership Void
A recurring theme in both the McKinsey manifesto and independent analyses of AI failures is the critical role of leadership. The manifesto notes unequivocally that there are no success stories where senior business leaders were not firmly in the driver’s seat. While IT departments provide essential support, the most successful “rewired” companies have business leaders actively conceptualizing, building, and running AI systems.
Yet, in practice, this kind of leadership is rare. Research indicates that six out of ten senior leaders admit their organization’s AI initiatives lack a clear owner. This leadership vacuum often results in siloed data, disconnected strategies, and a failure to tie AI investments to tangible business outcomes. The organizations that are breaking through this barrier are taking radical steps to align their leadership structures. For instance, some forward-thinking companies are merging human resources and IT functions, appointing leaders who can bridge the gap between digital transformation and workforce evolution.
McKinsey refers to this as the necessary “30-70 shift” in talent capability. Leading companies are ensuring that more than 70 percent of their tech talent is in-house, that 70 percent are hands-on engineers building solutions, and that 70 percent operate at a competent or expert level. As AI agents increasingly handle routine execution and coordination, human roles must shift up the value stack. Engineers must pivot from routine coding to designing complex architectures and guardrails, while business leaders must transition from task management to strategic objective-setting.
The Missing Measurement Layer
While McKinsey’s manifesto has been widely praised for its clear-eyed assessment of what it takes to succeed, independent analysts have pointed out a critical omission. Thiago Victorino, writing for the Victorino Group, published a comprehensive audit of the manifesto, praising it as McKinsey’s best piece on the subject to date. He particularly commended the inclusion of “digital trust” and “automating guardrails” as explicit themes, acknowledging that the excitement for agentic AI has outpaced many companies’ ability to manage the associated risks.
However, Victorino rightly argues that while ten of the twelve themes are directionally correct, they lack a crucial component: a governable measurement layer. A manifesto is a set of commitments, and commitments require a scoreboard.
“If you cannot measure humans and AI on the same scoreboard, you cannot govern either of them.” For example, McKinsey emphasizes the importance of increasing an organization’s “metabolic rate”—the speed at which a company moves from insight to decision, and from decision to action. While this is a powerful metaphor, it remains unauditable without specific metrics tracking cycle time, review time, and revision rates. Similarly, the mandate to build “digital trust” is presented as a perception outcome rather than a measurable infrastructure with defined thresholds and alarms. Without this integrated measurement layer, organizations risk managing their AI transformations through public relations rather than rigorous instrumentation.
The Path Forward
The convergence of McKinsey’s manifesto, KPMG’s ROI data, and independent audits paints a clear picture of the current enterprise AI landscape. We are moving past the phase of exploratory pilots and entering an era where industrialized value realization is expected. The companies that will thrive in this new reality are those that treat AI not as a software upgrade, but as a catalyst for comprehensive business transformation.
Leaders must stop asking what the technology can do and start asking what their organization is capable of absorbing. They must identify their core economic leverage points, demand that senior business leaders take ownership of the tech agenda, and build the necessary scoreboards to govern both human and machine performance.
The AI transformation manifesto is a valuable guidepost, but it is only the beginning. The real work lies in the messy, complex, and deeply human process of rewiring the enterprise for a future that is already here.