I have spent the better part of two decades inside large organisations, watching the same problem appear in different costumes.
A company scales quickly. Headcount doubles. New markets, new products, new reporting lines. Leadership hires smart people and builds processes to manage the growth. And then, somewhere in the middle of all that activity, something quietly breaks. Decisions slow down. Accountability blurs. Teams that should be coordinating are duplicating work or, worse, working at cross-purposes. The organisation is busy, but it is not moving.
I call it the moment when complexity outpaces execution. It is not a crisis. It is more insidious than that. Everything looks functional from the outside. The business is still running. But the operating model, the way work actually gets organised, decided, and delivered, has stopped serving the strategy it was built to support.
I have seen this at Visa, where I led regional product delivery across markets and global teams that each had their own history, their own ways of working, and their own ideas about how things should get done. I saw it at Mercer, running transformation programmes for business units navigating structural change. And I see it now, working with scaling companies and multinational divisions that are trying to grow without losing the ability to execute.
The typical response is piecemeal. A new governance framework here. A restructured team there. A process improvement initiative that solves one bottleneck and creates two more. These interventions are not wrong, but they treat symptoms. The underlying problem is systemic, and it requires systems-level thinking to address. That means understanding how the parts of an organisation connect, where the real friction lives, and what is actually driving the behaviour you are trying to change. Fix the wrong thing with confidence and you have just made the problem more entrenched.
What has changed recently is AI. And here is the take I do not hear often enough: agentic AI, deployed thoughtfully, does not just automate tasks. It can close operating model gaps that organisations have struggled to address through structural means alone. Coordination failures that stem from fragmented information flows. Decision bottlenecks caused by unclear accountability. Execution gaps at the seams between functions. These are not purely technology problems, but technology, applied with the right organisational understanding, can materially shift them.
The organisations that will benefit most from AI are not necessarily the ones moving fastest. They are the ones that understand their operating model well enough to know where AI can do genuine work, and where deploying it without that foundation will simply accelerate the dysfunction that already exists.
That is what I want to write about here. Operating model strategy, the systemic conditions that make AI transformation actually work, and the patterns I keep seeing in businesses trying to scale without losing coherence.