I went to Cannes half-expecting the future of marketing to argue with itself. The AI maximalists on one side, the craft purists on the other, everyone talking past each other.

That is not what happened.

Six stages. An agency, the biggest consumer goods company on the planet, a global marketer with Reddit, the people who measure advertising effectiveness for a living, a 150-year-old pharma giant, and the man trying to build artificial general intelligence. Six very different rooms. And the same conclusion kept walking back out the door.

The machine now does the volume. The human owns the inputs. Taste, truth, judgment, and creativity are the scarce thing, and they decide whether any of the volume is worth anything.

That sounds simple. It has teeth, because it cuts against most of what is being sold right now. Here is each room, what they said, and what I am taking back to clients.

R/GA's Tiffany Rolfe and Nick Pringle argued that when anyone can generate infinite content, the value moves up the stack. Away from the one-off asset, toward the system that produces it. "Intelligent brand systems." From making things, to making things that make things.

The clipped line was the framework. The line that moves money came in the close, almost as an aside: this turns what we do "not into a marketing expense, but actually an investment." For thirty years marketing has lived on the P&L as a cost, and costs get cut. Reframe it as a product that compounds and you change the conversation with the person who controls the budget. Their own number: great creative is 12 times more effective than mediocre, on the same media spend. We still book it as expense.

Full write-up: The R/GA line everyone missed at Cannes

The most disruptive company at Cannes was 188 years old. Marc Pritchard, P&G's Chief Brand Officer, showed a company that has rebuilt how the work moves through the building. From a batch process that took the better part of a year to a continuous flow measured in days. Every brand builder now works "as a direct-to-consumer founder with their hands on the keyboard." Agencies go modular. Retailers collapse into one ecosystem. In-house content factories run at 10x the old scale.

The receipts were real (Pantene live across Europe in months, 5x cheaper; Secret to +50% digital commerce; Mr Clean to 2x retail sales). The honest core kept it from being AI hype. Dawn's "clean as you go" product flopped in the UK because Brits soak their dishes, a cultural ritual no model would surface from millions of data points. A human caught it. The fix, "Skip the Soak," grew double digits. His close: "Robots don't build brands. You do."

My read: adaptability drives growth, and P&G just retired the "we are too big to move" excuse in front of the whole industry.

Full write-up: The biggest CPG can move fast. Your excuse just died

Shyam Venugopal, who runs global marketing and media transformation at PepsiCo, reframed the media question. Five years ago a meeting opened with "where are our ads running?" Now it opens with "how are our brands actually getting discovered?"

That changes what a platform is for. If the job is distribution, a platform is a channel. If the job is discovery, a platform is a signal system you listen to before you talk. PepsiCo mined Reddit for the taste insight behind Pepsi Zero Sugar and the flavour and texture insight that sent Doritos Protein back to R&D. He gave it a discipline, the 5S: how people search, scroll, screen, sense, and shop. And a global model: keep the spine consistent, let each market express it locally (the football platform travels, India makes it cricket).

The forward call is the one I keep thinking about. Discoverability gets much bigger "as humans start delegating more and more decisions to machines." For twenty years we optimised to be found by a person typing a query. Next comes being legible to an agent shopping on someone's behalf.

Full write-up: From "where are our ads running" to "how are we getting discovered"

The data session, and the one with the hardest numbers. Andrew Tindall (System1) and Eugene Healey (Studio EH) shared one of the largest creator-effectiveness studies yet: 64 million engagements, 5.4 billion impressions, eight markets.

Four things I took:

Healey's half was about building the creator like a brand: distinctive assets, a repeatable "social device," and the reminder that wear-out is a myth. Take his face out of a video and views drop tenfold, because the audience is hunting for the cue that promises a familiar experience.

Full write-up: Most creator marketing is a coin flip. Here is the playbook that isn't

One of the most distrusted industries on earth produced the most crafted, most trusted work of the week. In December 2025 Lilly became the world's first trillion-dollar pharma company, framed by CMO Lina Polimeni as "an overnight success 150 years in the making."

They did not invent a positioning. They interviewed their own people, from the executive committee to the person who seals the medicine box, and asked "what makes you proud to work here?" The truth was already inside the building: health above all. Founded in 1877 on the idea that the label has to match what is in the bottle.

The line I wrote down twice: a brand truth does not just tell you what to do, it tells you what to refuse. The briefs you will not take. A truth you cannot say no with is a slogan. And belief comes from action, not stories. Lilly defended diabetics' drug access at the Oscars in days, and ran free mobile mammograms at the WNBA All-Star Weekend, screening 120-plus women in a weekend.

Full write-up: An overnight success 150 years in the making

The loudest voice for restraint belonged to the person closest to building general intelligence. Demis Hassabis called AI "a bit over-hyped in the very near term" and still underappreciated over the next 10 to 15 years. The noise is loudest about the 12 months most likely to disappoint. Plan for the decade nobody is pricing in.

On what AI still cannot do: true creativity, which he described as a leap into the novel and new, as opposed to remixing what already exists. The CEO of a frontier lab, saying plainly that today's models remix and do not invent. On humans: "human intelligence is unbelievably efficient," in ways no model is close to. And the most relatable line of the week: his dream is to have his email automated. The man building general intelligence is bottlenecked by the same inbox you are.

Full write-up: The man building AGI told us to calm down

Two layers came out of the week.

The human layer first. R/GA, P&G, System1, and DeepMind are an agency, a CPG giant, a research house, and an AGI lab. They agree on the thing that matters. The machine handles volume. The scarce input is human: the insight, the taste, the truth, the judgment that decides which of the infinite options is any good. R/GA calls it the human input. Pritchard says robots don't build brands. System1 proves creativity beats every engagement metric five to one. Hassabis, of all people, says the leap is still ours. When those four land in the same place, it is not a trend. It is the ground.

Then the operating-model layer, which is where the work actually is. Speed (P&G shipping in days, not quarters). Discovery over distribution (PepsiCo treating platforms as signal systems, and the coming shift to being found by machines). Measurement (System1 killing the vanity metric). Brand truth as a filter for what to refuse (Lilly). And a posture of calm over urgency (Hassabis). None of that is about a tool. All of it is about how a brand is built and run.

The CPG read, since that is my world: the brands that win the next few years will treat creative as an asset that compounds rather than a cost to defend, reorganise around how they are discovered rather than where they advertise, measure brand memory rather than likes, lead briefs with fit and truth, and stay calm while everyone else over-rotates on the next model release.

Six rooms. One conclusion. The machine does the volume. You still own the part that matters.

Every country has different employment laws, payroll requirements, notice periods, and benefits expectations.

Oyster's Global Hiring Guides help you navigate the details, avoid surprises, and hire with confidence—wherever you're growing next.

Get the country guides