Marketing services will not look the same in five years. AI raises the ambition, not just the efficiency. Control every lever, and get paid on outcomes.

Pretorius wanted to be a musician but wasn't talented enough, so he studied law and loved it, ending up in media and telecoms law at Columbia in the mid-1990s just as the internet was being commercialised. He treated it as his ticket out of law, moved into digital media, and launched what he describes as the first ad network in South Africa. He then founded Acceleration, sold it to WPP some 12 years later, ran it there, helped then-CEO Mark Read turn around Wunderman, and became CTO in 2018. Acceleration was very technical ad tech and martech; the CTO role suddenly confronted him with the whole of marketing services. He built the group's open platform, and then AI arrived, making the last three years the most exciting of his career.

Everything you learn is interconnected. Pretorius says his telecommunications law background turned out to be extremely relevant, because network theory, resilience and node-based architectures are still fundamental principles today. It helped him write his own contracts as an entrepreneur and later understand the entirety of the problem, not just his specialism. WPP now talks about M-shaped workers, people who are both horizontal and deep in multiple areas, and by chance that is how his own career evolved. That versatility, he argues, is a real advantage when everything is moving fast.

Marketing has been technology-driven since the 1960s and 1970s, when Lester Wunderman used databases and segmentation algorithms in direct marketing. The internet accelerated it, and the big change now is that the medium itself changed the opportunities, through mobile and the social economy. The old model of media companies creating content to attract eyeballs and sell advertising has fundamentally changed. Agencies should not be nostalgic about any channel; they need to follow the eyeballs. Executing in one channel is easy, but allocating across channels and understanding how they combine is far harder, and that complexity is what keeps clients coming.

Pretorius spends 40 to 50% of his time with clients, on the principle that you understand nothing except through the lens of what matters to them. Over the last 10 to 15 years, more clients want to own the core intelligence, the data and insights that drive their marketing and their spending decisions. Building commercial data platforms inside the organisation is now the main thing his clients are working on.

There is a philosophical split. WPP believes in data collaboration, distributed intelligence, and that clients own their consumer data. Others believe they can be gatekeepers and intermediaries who lock clients into their data asset. Pretorius quite likes the divergence, because it gives clients clear commercial optionality. It is very boring, he says, if everyone is saying the same thing.

For the first time, Pretorius has had to think not just about how technology lets the agency deliver differently, but about how it changes the nature of work. AI is changing how consumers search, discover and buy; it is changing production most sharply in the short term, making downstream studios, lighting crews and camera people less necessary; and it is changing how teams work.

WPP Open integrates the end-to-end supply chain, from strategy to creative to production to media to CRM, on one data layer, and then agentify processes. Sometimes work is fully automated; more often humans orchestrate workflows and bring in specialist agents. Pretorius calls it agents in the loop, not humans in the loop. Instead of five agencies doing a project, five specialists work with a bunch of agents on Open, a speedboat model that gets the work done.

Efficiency is the boring story that will happen anyway. There are tasks humans should never have done: ad trafficking, in his words, is a soulless, life-destroying exercise, vital to get right but mechanical and automatable. The excitement lies elsewhere.

In strategy, WPP built composite complex strategy agents that do about two weeks of work in five minutes, surfacing not just the obvious answer but the non-obvious one, work no client would once have paid for. In creative, when Veo 3 launched they made a football-through-the-ages film for Google, about 15 AI-generated scenes that would have needed 15 locations and a feature-film budget, made in under a week with three thousand prompts. The quality is now so high that if you can tell it is AI, it was used badly, and the negativity about AI slop is a slightly boring meme.

Clients moved from wanting AI to make work cheaper, to wanting more content for the same money, to wanting better content that is tested to make sure it works. WPP has reset its unit economics and keeps making its models more aggressive. Because it can control the whole supply chain, many of its largest new models and recent pitch wins are partly or entirely output based: it gets paid when the client sells a product. That only works when one partner controls every lever, and more sophisticated clients now see that end-to-end control can drive commercial innovation.

Adoption and change management is the hardest part of any transformation. WPP has been structured about it: a Hex creative technology apprenticeship hiring AI-native talent from diverse backgrounds, aiming at a thousand apprentices by the end of 2028; seven years of AI training across the workforce since 2019; a bought-in research lab and a spun-out deep research lab; and executive courses at Oxford Said. But the most powerful thing, Pretorius says, is distributed innovation: you empower people with general purpose tools, put guardrails around them, and let them free. A no-code agent builder launched at the end of 2024 produced thirty thousand agents within a month, and the best super-agent ideas came from the community. That bottom-up energy drove the transformation more than anything.

Partnerships, built on people, platform and partners, have become existential and always on, but also more fluid: pricing models go out of date within months when volumes come in ten times higher than predicted, leading to almost ridiculous conversations about cutting prices by 90%. On governance, Pretorius works from first principles: never violate copyright, never share client data without consent, never use client data to train models. Partnered with general counsel Vicky Brown from day one, he is confident that what WPP builds is legal and safe by design, because client data security and copyright liabilities are sacrosanct.

WPP's new focus, announced by CEO Cindy Rose, is being the trusted growth partner for brands in the age of AI, a vision aligned with client success. That can mean commerce platforms, influencer programs, data analytics and CRM, so what the group sells will be very different. The existential question is software and human work: teams tell Pretorius their apps are writing themselves and mean it, and people are shipping forty thousand lines of code a day. If anyone who can specify a requirement can spin up software of any complexity, it raises open questions about productivity, differentiation and trust. He thinks WPP's mix of creativity, data, clients and domain expertise creates moats, but expects a completely wild ride.