As the head of technology across EMEA — one of the most diverse regions globally in terms of digital and AI maturity — how do you navigate such variation, from highly regulated European markets to fast-moving Gulf economies and emerging African ecosystems?
In many ways, it mirrors what we saw during the early phases of cloud adoption. Different organisations and countries were all at very different stages of maturity. To make effective use of cloud — and now AI — it’s not just about having an ambition or end goal. It’s also about addressing skills and organisational readiness.
My teams across Europe, the Middle East, and Africa work backwards from the customer’s needs. That includes regulatory requirements as well as the organisation’s internal maturity. For customers migrating and modernising, skills development is a critical part of the journey — not just for where they are today, but for where they want to be tomorrow. That’s why building long-term capabilities within regions, particularly around technology and AI skills, remains so important.
What separates organisations that are genuinely transforming their operating models from those still stuck in migration mode?
It comes down to fundamentals that have always shaped technology adoption. The first is organisational will — a genuine desire to change. That has to be driven top-down. The second is having a clear goal, and the third is having the skills to support that ambition.
Where organisations struggle is often at the leadership level. If the board isn’t aligned on change, it becomes very difficult for teams lower down to drive meaningful transformation. In contrast, organisations that are more mature typically started earlier and had a stronger appetite for change from the outset.
Even with AI and digital transformation now firmly on board agendas, do you still see hesitation or uncertainty among organisations?
Yes, in some cases — and often it comes down to not knowing where to start. That uncertainty can make transformation feel more complex or risky than it needs to be.
What we focus on is helping customers move from A to B while demonstrating tangible progress along the way. If the only goal is set several years out, with nothing delivered in between, organisations can lose momentum.
Where I see real progress is with organisations that are willing to reimagine their processes. A good example is TP ICap, a global brokerage with a business unit called Parameta. Their products must comply with local regulations in every country they operate in. Previously, manual compliance checks could take up to a month. With a generative AI solution, that process now takes significantly less than a day.
That kind of outcome requires willingness — a decision to use technology to achieve a clear business result. It also requires focus. Rather than running dozens of proofs of concept, the organisations that succeed select a small number that are genuinely impactful and move them into production.
Do you see these differences more at a country level or an organisational level, particularly?
I don’t really see markets lagging behind. What varies more is organisational ambition and the timelines leaders set for achieving it. Every organisation has ambition, but how quickly they want to move depends on leadership — the CEO, the board, and the level of local sponsorship within the organisation.
re:Invent 2025 saw major announcements around the Amazon Nova model family. How do model choice and flexibility influence how organisations build AI into their products and services?
Choice has always been central to how Amazon approaches AI. Different organisations have different cost sensitivities, data profiles, and use cases, and that flexibility is critical as enterprises move from experimentation to production.
Models such as Nova Lite address customers with straightforward text-to-text or video requirements, offering an accessible entry point. At the other end of the spectrum, Nova Forge is designed for organisations managing large and complex datasets. By allowing customers to bring their data into the model lifecycle earlier, it reduces the time required to develop a model that is appropriate for their specific use cases.
For enterprises dealing with regulatory complexity or operating across multiple markets, that acceleration matters. The ability to work with data earlier in the process helps shorten development cycles and move AI initiatives into practical use faster, particularly for organisations focused on operational outcomes rather than isolated experimentation.
Frontier agents are increasingly part of the enterprise AI conversation. How do you see their role evolving?
Frontier agents bring scalability and autonomy — the ability to operate independently for extended periods. But what’s equally important are the controls around them.
Identity, credential management, and policy are critical. Organisations need visibility into what agents are doing alongside humans, just as they did when cloud adoption raised early questions around security and resilience. Observability and governance frameworks are essential to ensure trust and accountability.
I also come at this as a former developer — I started my career working with COBOL, Fortran, and Pascal. Many organisations still rely on applications that are no longer fit for purpose. Accelerating modernisation is essential, particularly where technical debt or licensing constraints have slowed progress. In that context, automation, DevOps agents, and security agents can make a meaningful difference.
Looking ahead, how do you see enterprise transformation evolving across the region?
There will continue to be rapid evolution around agents and compute, particularly as organisations look to scale AI responsibly. Beyond individual technologies, the bigger shift is in how enterprises reimagine end-to-end processes using agentic AI.
That work is already underway with customers across Europe and beyond. As capabilities mature, quantum will also enter the conversation — especially for organisations that require enormous computing power, such as those using high-performance computing.
Over time, I see a future where working with a mix of humans and agents becomes a normal operating model for enterprises.
Discussion about this post