McKinsey has just told on itself. In a July 2026 instalment of its Rewired series, senior partner Brooke Weddle admits what every consultancy watching AI transformations up close has quietly known for a year: most organisations are stuck. Not because the technology does not work, but because the organisation around it has not changed shape.
This matters because Rewired is the closest thing management consulting has to an official AI transformation manual - built from more than 200 enterprise-wide technology and AI transformations, tested and re-tested. When its own authors start conceding that the hard part was never the model, it is a signal worth taking seriously. It confirms something we have been arguing from the African mid-market for two years: becoming an AI-oriented enterprise is an organisational redesign problem wearing a technology costume.
The organisation around the technology has to change shape.
Quick answer
An organisation moves from AI activity to enterprise value when it focuses on one to three domains, coordinates decentralised experimentation through a cross-functional nerve centre, redesigns roles and workflows, and gives people room to learn in public. At Green Everest, that organisational destination is the Adaptive Intelligence Organisation: a business redesigned to capture measurable value from AI, not merely retooled with it.
The activity trap
Most companies today have done something with AI. They have run pilots, deployed a handful of use cases, and generated a burst of internal excitement. Weddle describes a professional-services CEO proud of the agents his firm had built to analyse prior wins and losses - until he recognised the deeper problem: no domain map, no clear way to measure progress, and too many initiatives running in parallel with none of them compounding.
That pattern is close to universal. Activity is not the same as value, and most organisations cannot yet tell the difference because nothing forces the discipline of choosing. The fix is not more pilots. It is fewer, sharper ones. McKinsey's Rewired research finds that the organisations that scale most successfully focus on one to three domains where AI can transform performance, then build for scale from day one. The real cost is not the first deployment. It is maintaining and scaling it once the excitement fades. That is why CFOs are increasingly in the room from the start, not brought in later to sign off.
The board test
If you cannot name the one or two domains where AI has to work, you do not have a strategy. You have an experiment budget.
The missing nerve centre
The second gap is structural. Most organisations do not want, and should not build, a heavily top-down AI programme. AI adoption is inherently decentralised: the people closest to the work are the ones who find the real use cases, and they resist being told exactly how to use the tools. But decentralised without coordination becomes chaos.
What the best organisations are building instead is what Weddle calls a nerve centre - not a control tower, but a coordinating layer that gathers signals from across the business, processes them, and helps successful approaches spread. In practice, that is a standing coalition of business, technology, finance, HR, and operations leadership, working from common metrics and a shared, plain definition of what success looks like for every initiative. It is the layer that turns forty scattered experiments into one organisation that learns.
Most companies do not have this layer. They have a technology team running a backlog and a leadership team receiving updates. Neither is a nerve centre. Building one - properly resourced, cross-functional, with real authority over what gets scaled and what gets stopped - is now one of the most consequential moves available to a board.
Leadership as the actual bottleneck
Here is the uncomfortable finding: the constraint on AI transformation is rarely technical any more. It is behavioural. Employees are managing genuine optimism about what AI makes possible against genuine anxiety about upskilling, job security, and an unstable external environment. Leaders who pretend the anxiety is not there lose the workforce. Leaders who name it directly, while painting a credible picture of what becomes possible, keep it.
The organisations pulling ahead are not the ones with the best models. They are the ones actively building space for learning: capturing what worked and what did not, telling the stories, and rewarding people for trying things in public rather than punishing visible failures. That requires a specific kind of leadership behaviour - inviting friction rather than suppressing it, and treating courage and continuous learning as core competencies rather than nice-to-haves. This is not soft-skills window dressing. It is a practical differentiator between organisations that scale AI and those that stall at pilot.
Culture compounds - or it does not
There is a deeper point buried in the Rewired findings, and it is the one that should worry every leadership team most: speed is no longer purely an AI phenomenon. It has become a defining organisational capability in its own right, independent of which tools you deploy. If culture is the values, behaviours, and mindsets that shape how people work, then AI transformation is not a parallel workstream to culture. It is a culture workstream, executed at the level of the individual role.
The clearest illustration is a persona-level transformation, not an enterprise-wide one: frontline workers with decades of tenure and real pride in their craft, whose job changes from escalating a problem to a supervisor to directing a set of AI-generated options themselves. The technology was not the hard part. Rebuilding trust, status, and social interaction around a changed role was the actual transformation.
The real headline
Strip away the case studies and one sentence carries the whole argument: an AI transformation is a people transformation. The organisations that will win this decade are not the ones with the most advanced models. They are the ones fastest to understand how their people reach full potential inside an AI-enabled way of working - and disciplined enough to build the domain focus, the nerve centre, and the leadership behaviours that make that possible.
That is what it means to become an AI-oriented enterprise: not an AI-adopting one collecting use cases, but one that has rewired its structure, coordination layer, and leadership culture around the technology - deliberately, and in that order. In Green Everest's architecture, this is the move towards an Adaptive Intelligence Organisation: strategy, people, workflows, governance, and measurement redesigned as one system.
The technology was never the constraint. It still is not. What has changed is that one of the world's largest management consultancies is now saying so in print.
FAQ
What is an AI-oriented enterprise?
An AI-oriented enterprise redesigns its domains, workflows, roles, coordination, and leadership practices around AI. It does not treat AI as a collection of isolated tools or pilots.
What is an AI nerve centre?
An AI nerve centre is a cross-functional coordinating layer of business, finance, HR, technology, and operations leaders. It connects decentralised experiments, establishes common measures, and decides what to scale, stop, or spread.
Why should organisations focus on one to three AI domains?
Concentrating on a few high-value domains gives an organisation enough focus to redesign workflows, build the operating model, measure economics, and sustain adoption. A wide portfolio of disconnected pilots tends to spread activity faster than learning.
Why is AI transformation a people transformation?
AI changes how decisions are made, how roles create value, and how people learn and coordinate. Capturing value therefore depends on leadership behaviour, trust, capability building, and role redesign as much as it depends on the technology.
Can your leadership team name the domain where AI has to work, the number it must move, and the person who owns the result? Start with the AIO Quick Scan. In ten minutes, it will help you see your current position, surface the binding constraint, and choose a practical next move.
Sources: Brooke Weddle, Rewired takes: Practical people lessons for scaling AI adoption, McKinsey & Company, 13 July 2026; McKinsey & Company, Rewired in action.