15 minutes with: Derreck van Gelderen

What brought you to PA?

Before joining PA, I built my foundations in engineering and data science. I’m half-French, half-Dutch, and moved to the UK when I was 12. I studied mechanical engineering at the University of Bristol, followed by a PhD in probabilistic fracture mechanics, which is where I really got into coding, software development, and machine learning.

I began my career working across the nuclear sector with clients like EDF Energy and Hitachi-GE. It was a great way of building deep sector knowledge and developing my technical skills. But over time, I wanted to take a more active role in shaping innovation, and increasingly felt the limitations of trying to drive change from within.

That’s what brought me to PA. Over the past six years, I’ve had the opportunity to apply innovation across some of the UK’s most complex and critical sectors, from nuclear and energy to healthcare, consumer, manufacturing, and government.

My very first project at PA was in healthcare, where I developed activity-based job planning models to forecast how capacity could shift to align with high-demand services. I later worked with the Department for Business and Trade to establish a Data Science Innovation Lab, where we deployed AI solutions to accelerate specialised manual processes. During the COVID-19 pandemic, I served as Chief of Staff to the Head of Tech and Data on the national vaccination programme, helping to deliver the data and technology infrastructure to record up to five million vaccinations per week.

In recent years, my work centres on helping leaders turn AI into meaningful value. This ranges from securing board-level investment, through to advising private equity and infrastructure funds on AI-led value creation, diligence and data centre strategy, all the way to developing first-of-a-kind AI solutions.

This blend of hands-on AI development with leading large-scale, strategic programmes has fuelled my time at PA. It’s enabled me to anchor my advisory work in deep practical experience, and to help clients achieve outcomes they hadn’t thought possible.

How would you describe your role to someone you’d never met before?

I’d say I do two things.

First, I help clients answer business-critical questions: What does AI actually mean for our organisation? Where should we start? How do we quantify the value? What skills are needed to deploy and embed this technology effectively? Second, I build AI solutions that don’t yet exist in the market, whether that’s pioneering an industry-first application or designing a bespoke solution to solve a complex challenge no off-the-shelf product can address.

I lead our cross-firm AI advisory capability, and our AI work in the Energy and Utilities sector. My role blends strategic and hands-on: I typically engage with clients every week across projects involving AI, data, digital transformation, and business strategy. What I enjoy most is working directly with leadership teams to shape the opportunity, and then helping to build something truly new. For example, we developed the UK nuclear decommissioning industry’s first Gen AI-powered digital colleague with the Nuclear Decommissioning Authority and Sellafield. This work built on our earlier solution, SLComply.ai, which reduced engineers’ admin burden by 90 percent. Our project won ‘Best Example of Applying Creative and Innovative Solutions’ at the NDA Group Supply Chain Awards 2024, and ‘Best Use of AI in Energy and Utilities’ at the AI World Series 2025.

What makes PA different?

During my interview to join PA, a colleague said: “If you have a great idea here, people will rally behind you to make it happen.” That’s exactly what I’ve found. Good ideas don’t need to come from the top. PA has a distinctly entrepreneurial energy. You’re encouraged to lead, experiment, and create, not just deliver.

Our first AI solution for Sellafield began as a lunchtime chat. A colleague and I mocked up a quick demo to bring the concept to life. A few conversations later, we were presenting it to clients. The culture and ecosystem at PA make it possible to turn ideas into impact, fast.

What sets us apart is how we’ve consistently evolved with both our clients and the technology. Two years ago, our work centred on identifying practical AI use cases, and shaping foundational blueprints to guide adoption. A year later, the focus shifted to embedding early Gen AI solutions, and building targeted products to test real-world impact in controlled environments.

In our interactions with clients, our advice is grounded in real-world delivery. We mobilise multidisciplinary teams, not just to consult, but to create, without the layers of complexity in much larger firms. That’s what enables us to deliver results that are both credible and new.

What are the biggest challenges your clients face?

The overriding challenge for our clients is managing the strategic tension between the pace of technological change and their internal appetite for progress versus risk.

The single biggest barrier comes from organisations driving with one foot on the accelerator and one foot on the brake. Most boards and management teams happily claim they’re bold while behaving cautiously, which calcifies AI into a permanent nice-to-have rather than a core strategic capability. The result is often a misalignment between stated strategic ambition and genuine resource appetite.

Challenges manifest differently across domains. In defence and security and heavily regulated sectors, data sovereignty and hardware constraints tend to dominate. In other organisations, the issue is often a lack of in-house technical expertise, whereas for more fast-moving sectors like consumer and entertainment, competition, and speed are key factors driving the urgency. A particularly acute challenge in the nuclear decommissioning sector is strategic workforce planning and knowledge retention. The impending retirement of up to 20 percent of the workforce with decades of experience is creating a critical knowledge transfer gap where Gen AI and Agentic AI has a clear role to play.

Data readiness is another key barrier. For two decades, organisations chased the myth of perfectly governed data. This has created a strange tension today. Everyone’s excited about AI agents coordinating whole workflows, but insist “our data isn't ready yet.” This means they can end up kicking the can down the road. High performers are flipping this equation. Instead of spending years ‘tidying the whole house’ with broad data programmes, they focus on targeted, high-value workflows. The question shifts from “When will our data be ready?” to “What targeted data improvements do we need so this specific AI use case works reliably?”

Importantly, a key challenge across organisations is the overstatement of AI adoption. Most have spent the past two years doing pilots, sandboxes, and a fair amount of theatre. Products like Copilot have been rolled out to thousands of employees, helping people draft emails and documents a bit faster, but this is nowhere near the kind of universal, habitual usage we’ve seen on the consumer side with ChatGPT. But inside companies, these solutions are still treated as ‘productivity of convenience’ – optional accelerators that sit on the edge of real work, so behaviour and core processes haven’t fundamentally shifted.

Finally, there’s landing on the right path to production. Two years ago, most organisations were still wrestling with the question of where to use AI. The challenge has now fundamentally matured.

For high-performing organisations, the questions of tomorrow won’t focus on tools. They’ll focus on measurable business impact: “What percent of my cost base is now touched by AI every day?”, “Which decisions are measurably better or faster because of AI?”

How do you solve those challenges in ingenious ways?

Clients can’t start with an AI strategy in isolation. Instead, they need to anchor everything around the business strategy and identify where AI can actually drive that strategy forward. That’s why we use an ‘AI blueprint’ as the first step, because it’s all about understanding the real business problems and the unmet needs, not just the technology. This opens up a broader field of possibilities and avoids limiting solutions to their current understanding of AI.

Within AI, strategy and execution are blending. This means that our clients don’t want to see just paper after working with us for eight weeks, they want to see something real, as we’re doing it. So, a lot of the projects we deliver are about bringing very different skill sets together to build something tangible with the client in months, not years, so they see real results quickly. It also means that our advisory work is anchored in practical experience of building solutions for our clients, instead of just handing over a plan and walking away.

Alongside this, we put as much emphasis on responsible adoption and enablement as on the technology itself. We help clients assure that their AI systems are fair, transparent, and accountable, building governance and reporting that, in time, will sit in an annual report. Enablement is about helping people relearn how to do their jobs with AI embedded in workflows, not about teaching them how to use a new app.

In the last 12 months, we’ve seen increasing requests to help develop Agentic factories. This is about scaling up and creating a permanent capability where human and AI agents work together. We keep a user-centred mindset from the start, designing systems that are intuitive for users, with the flexibility to pivot as technology and pricing evolve.

Which projects are you most proud of?

It’s genuinely hard to choose. Each project has been a valuable learning experience, teaching me something new, whether from the technology we used, the incredible clients, or the talented teams I worked alongside.

That said, one that stands out is delivering SLComply.AI for Sellafield. It started as just an idea, born out of seeing untapped potential in the nuclear sector back when the key terms to know were machine learning and natural language processing.

It’s a project I’m particularly proud of, not just because it won awards, but because it showed how we could fill a real gap and pave the way for more innovation in the industry. For example, it led to the co-creation of DANI2, the UK’s first nuclear decommissioning digital colleague powered by Gen AI. And, it’s a reminder that the best work is always a team effort.

Another stand-out project was the COVID-19 UK Vaccine Taskforce. I was fortunate to be able to run workstreams to increase the number of people who were vaccinated each week. Everything rallied around that single objective, and you could really see the impact in the real world. I also learned a lot about leading teams and large programmes in an environment where every second counts.

The role we play always depends on the client’s situation and circumstances. If you have a confident client with lots of enthusiasm to innovate, you can build really exciting things. For instance, we worked with a large OEM to pioneer something they didn’t think could be done. In just three months, we delivered the first version of a near real-time emissions tracking platform, enabling global monitoring of sustainable aviation fuel (SAF) uptake.

Recently, we’ve been working with a leading sports entertainment company to shape their Data and AI blueprint and support them with the first stages of implementation. This one is personally meaningful to me because their brands were a big part of my childhood. Getting the chance to work with the organisation behind the shows I watched growing up has been a bit surreal in the best way. It’s been incredibly rewarding to help them with both advisory and execution. It’s a great example of how combining deep industry knowledge, modern data foundations, and Gen AI can open up completely new ways to operate and engage fans on a global scale.

What advice would you give to someone who wanted to follow in your footsteps and do what you do?

I didn’t follow a grand master plan. I just followed my curiosity, said yes to interesting problems, and learned a lot through trial and error. What worked for me won’t be necessarily a perfect template for anyone else. The important thing isn’t the specific steps, but developing the mindset: being curious, comfortable with ambiguity, and willing to keep learning.

People often say, “do what you love and it won’t feel like work.” I’d put it slightly differently. If you care about what you do, you’ll have more energy for it. In this type of role, there’s always another challenge to take on or idea to explore, which is part of the fun. The flip side is that the work can easily expand to fill all the available space, so it’s important to be intentional about where you spend your time and what you say yes to. That’s how you keep it rewarding rather than draining.

What’s made the biggest difference for me hasn’t been some secret technique or clever framework, it’s been the right mentors. The most valuable things I’ve learned are about dealing with uncertainty, managing pressure, and working through issues when things don’t go to plan. That’s the stuff you pick up from watching how good people operate, not from a textbook.

And finally, don’t try to do it all on your own. This work is much better, and usually much more successful, when it’s done in teams. There’s real strength in collaboration: you move faster, you make better decisions, and you actually enjoy the journey. If you try to carry everything yourself, you slow things down and miss out on what others can bring.

What are your future goals, professionally and personally?

As we move into 2026, I look forward to continuing to partner with our clients by co-creating and scaling ingenious solutions using AI and Agentic systems that can deliver a lasting impact.

In parallel to helping PA clients, I’m also really excited about continuing to transform my parents’ business by integrating Gen AI and AI Agents into it. I want to help them move into a space where AI agents can work alongside them and really modernise how they operate. It’s a personal project that I’m really passionate about and looking forward to seeing take shape over the next few months.

And on the personal side, I used to be really into rugby and judo, and coached judo for a while until an ankle fracture put me on the bench. Now I’ve hung up the boots and the gi, I’m really enjoying getting back into tennis. I have to admit, I’ve caught the padel bug too, so if anyone wants to have a meeting on the court, I’m all for it.