The companies that are pulling ahead on AI are not the ones with the most pilots. They are the ones that have understood, often painfully, that scaling AI is an operating-model exercise more than it is a technology exercise. The technology is necessary but nowhere near sufficient. What separates the winners is what they do around the technology — how they organise, decide, govern, hire, and learn.
Rewiring the operating model is harder than buying a platform. It is also where the durable advantage lives.
The four dimensions of rewiring
When we work with organisations on this, we look at four interconnected dimensions. Each can be tackled on its own, but the compound advantage comes from working them together.
1. Strategy — what is AI for in this business?
Most AI strategies are really just lists of use cases. A real strategy answers the prior question: where will AI compound advantage for us, and where will it not? Which P&L lines do we expect to move? What is our point of view on build-vs-buy, on data sovereignty, on model providers? Without that, every use-case prioritisation becomes a negotiation between teams with different assumptions.
2. Foundations — tech and data that can carry the load
Scale needs plumbing. Identity, integration, observability, evaluation, data catalogues, model governance. The organisations that did the unglamorous foundation work early are now deploying their fifth and sixth use cases in weeks rather than quarters. The ones that didn't are still relitigating their data architecture every time a new agent ships.
3. Operating model — the workflow itself
This is where most programmes stall. You cannot drop AI into a workflow without changing the workflow. Hand-offs move. Decision rights shift. SLAs change. New roles appear (AI ops, prompt engineering, agent supervision); old roles consolidate. Most organisations underestimate this and end up with AI that runs next to the business rather than inside it.
4. Workforce — augmented, transitioned, supported
An AI-augmented workforce is not a thinner version of today's workforce. It is a differently-shaped one, with different skills, different tools and a different daily rhythm. Rewiring the operating model means rewriting the role descriptions, the career paths, and the learning programmes that support them. This is the dimension where leadership credibility is won or lost.
AI does not transform an organisation. The organisation transforms itself, with AI in the blast radius. Pretending otherwise is the most expensive mistake in the market right now.
What "rewired" looks like in practice
A rewired operating model has a few visible markers:
- Workflows are designed for AI and human collaboration from the start, not retrofitted around an existing process map.
- Decision rights are explicit — what the AI can do alone, what needs a human, what needs an executive.
- Data flows are continuous, not batched. Models learn from real usage, with the right feedback loops in place.
- Governance is integrated into delivery, not a separate gate that slows everything down.
- Workforce changes are planned in waves, with transition support that is funded and visible.
None of this is exotic. Most of it is good operating practice. What is new is the scale at which it has to be done, and the speed at which the underlying technology is evolving underneath it.
Why this is a leadership problem
Rewiring an operating model is not delegable. It cuts across functions, budgets, hiring plans and external commitments. It requires explicit air cover from the top, sustained over multiple quarters. The organisations that succeed treat AI as an executive agenda item — not because the CEO needs to choose models, but because only the CEO can resolve the trade-offs that AI surfaces.
The CFO has to reallocate capital. The CHRO has to redesign the workforce plan. The CIO has to make platform calls that lock in optionality. The COO has to redesign the operating cadence. None of these happen at the AI council on a Friday afternoon. They happen when the leadership team has internalised that this is an operating-model rewire, not a tech rollout.
— the takeaway.
From pilots to platforms to plumbing
The next chapter of AI for most organisations is not more pilots. It is fewer, deeper, and integrated into a deliberately rewired operating model. That is harder, slower at the start, and far more valuable at the end.
The companies that do this work in 2026 will compound on it for a decade. The ones that keep running pilots will spend the same money and still be in proof-of-concept territory in 2030.