The State of Transformation and AI Adoption

The gap is not technology

The Middle East is all-in on AI. The ambition is funded, pilots are everywhere, and people here use AI at work more than the global average. AI could add around $320bn, close to 11% of regional GDP, by 2030.

Yet far less of that turns into real value, and the reason is not the technology. Most transformations miss their objectives, and AI is only the clearest example of a much older pattern. The constant is never the tool.

It is everything around it that was never changed: the strategy, the data, the process, the people and the controls. Leaders use AI as the reason to change how work is done. The rest add it on top of a business no one redesigned, so little really changes.

We have seen this wave before

Every wave of technology promised transformation and delivered a fraction. Service-oriented architecture promised reuse and left messy processes and data.

Cloud promised cheaper, faster IT and too often hosted the same processes at a higher bill. Automation promised hands-free efficiency and bolted bots onto work no one had fixed. AI is the newest wave, not a new pattern.

The problem compounds. Each wave is added on top of the last unsolved problem, so by the time AI arrives the gap has multiplied rather than reset. The technology changed every decade. The failure mode did not, because success is roughly four-fifths change and one-fifth technology.

The foundations you already paid for are underused

Before AI even arrives, value is already leaking from the platforms, cloud, data and people you have invested in. About a third of ERP capability typically sits unused after go-live.

Around 29% of cloud spend is wasted, and most organisations struggle to manage it. Data readiness is the single most cited obstacle to AI. And transformation is four-fifths change, where resistance, not technology, is the number-one barrier.

This is the real starting point. Not a shiny new platform, but the gap between what you already own and what you actually use.

Do not migrate the problem

This is why a platform migration is not a transformation. Replace a CRM without redesigning the process, unifying the data and changing how people sell, and you rebuild the old mess in a newer tool: the same shadow spreadsheets, the same low adoption, a bigger bill.

Lift and shift to the cloud without redesign, and you pay more to run the same inefficiency. Bolt AI onto work no one redesigned, and you only automate the mess faster.

The honest counsel is uncomfortable: if you will not fix the process, the data, the adoption and the governance first, do not migrate yet. The platform was never the problem, and it will not be the solution.

Six dimensions decide whether value is realised

We read transformation and AI maturity on six dimensions: strategy and value, data and AI, platform and architecture, process and operating model, change and adoption, and governance and risk.

Together they are the necessary and sufficient set that decides whether value is realised. Weak in any one, and value leaks. Strategy and change carry the most weight, because the gap is human and operational far more often than it is technical.

The recipe: fix just in time, not for five years

The most expensive myth in enterprise transformation is that you must fix every foundation before you can begin. It sends organisations into multi-year programmes that bank no value and quietly exhaust the appetite for change.

The recipe is the opposite. You do not wait. You diagnose where value leaks, anchor on the one outcome worth the most, and fix only the slice that outcome needs: the necessary and sufficient cut of process, data and adoption.

You redesign the work rather than replicate it, so you stop carrying the problem forward, and you drive real adoption, which means changing how people work, not running the ceremonies.

You govern by design, and you run it and bank the value. First value in a quarter, not years, and the first win funds the next. You run it at two speeds: the slice delivers value now, while AI compresses the wider foundation, data engineering, migration, process mining and testing, so the full fix takes months, not years.

The Middle East difference: govern by design

In the Gulf, where data sits and who can touch it is set by law. Saudi Arabia's PDPL, enforced by SDAIA, keeps personal data in the Kingdom unless transfer is approved. The UAE's PDPL and sector rules keep health, finance and government data in country. Public-sector and regulated data must stay on licensed, in-country cloud.

So governance and sovereignty cannot be an afterthought. Treat residency, security and approvals as design inputs, choose sovereign or in-country cloud for regulated workloads, and classify data early so AI can use it without breaking the rules. Here, governance is built in by design, or rebuilt later at a far higher cost.

What changes with agentic AI

The next wave is already arriving, and it raises the stakes. Earlier technology informed a decision or automated a defined task. Agentic AI takes the action. Pointed at a sound process, it compresses work that took weeks into hours. Pointed at a broken one, it executes the broken process at machine speed.

That is why the foundations matter more, not less. An agent inherits the quality of the data, the process and the controls beneath it, then acts on them autonomously.

The discipline is the same as the rest of this argument, applied with less margin for error: clean inputs, clear boundaries, an audit trail, and a value case before scale.

And on anything safety-critical or in operational technology, a person stays in command of the action, by design. Agentic AI is not a reason to skip the foundations. It is the reason to get them right.

Diagnosis, prognosis, prescription

The way through is clinical. Diagnose where you stand on the six dimensions. Read the prognosis honestly: left unchanged, the gap compounds and your share of the value migrates to whoever moves first.

Then prescribe the fix, just in time and sized to the next outcome. That is exactly what the Value Gap assessment does, in about ten minutes.