Published July 14
As AI dominated conversations at this year’s Cannes Lions International Festival of Creativity, the same questions surfaced again and again: Where will it create the biggest impact? And what does it mean for the future of creative work?
For Amazon Web Services CMO Julia White, the answer lies in AI agents. Over the past year, she has pushed AWS to become “the most AI-forward marketing team on the planet,” reimagining everything from localization to web publishing through an agent-first lens.
The goal wasn’t simply to move faster. It was to eliminate repetitive work so marketers could spend more time doing what AI can’t: understanding customers, crafting better stories for clients and generating stronger ideas. Today, White estimates that AWS marketing has deployed thousands of AI agents across the organization.
Speaking with The Current at Cannes Lions, White discussed what AWS has learned, why every CMO needs to get hands-on with AI and why the industry’s biggest challenge isn’t creating more content, but creating better ideas.
Last year you challenged your team to become “the most AI-forward marketing team on the planet.” What led to this approach?
When you think about the AWS brand, we’re a provider of AI technology to the world. If our marketing isn’t representing that, we’re not very good brand ambassadors. We have all of the capabilities in-house. I was like, “We have no excuses.”
We really started by just getting people comfortable with AI. Everyone was encouraged to go try it, build an agent, see what they could do. That first phase ended up creating a lot of duplicative tools and approaches, but it allowed us to learn and mature into the next phase.
We then set up five workstreams focused on core marketing workflows that we knew agents would transform. Each team included marketers and technologists working together in two-week sprints. Our mental model was build a skateboard. If that works, build a scooter. If that works, build a car. But if the skateboard doesn’t work, throw it out. Learn quickly and iterate before going all in.
What kinds of efficiencies have you seen over the past year?
Localization is a great example. Linguists often had to rewrite 80% of the content just to make it sound natural. Now we inject an agent between machine translation and the linguist, using local colloquialisms and improving grammar. That means the linguists spend their time doing what they’re actually trained to do. Previously, translations could take up to two weeks. Now they’re typically completed in no more than a couple of days.
The biggest gains have been in content. If I take the example of creating new webpages, which we do around 10,000 of every year, the process used to involve copying content from different systems, checking templates, testing links and publishing. It took about four to four-and-a-half hours per page. When we stepped back and added agents with the human reviewing, it dropped to about 10 minutes. That’s roughly a 95% improvement.
Beyond productivity gains, how has AI changed the role of marketers?
There’s a strong narrative around job replacement, but what we’re actually seeing is AI freeing our linguists and our marketing talent to do what they were trained to do instead of the repetitive parts of the work. One of the early things we did was ask people to write down their “paper cuts” — the things they hated most about their jobs. Some of our AI teams simply went after those paper cuts. Once people saw those repetitive tasks disappear, they were delighted.
Smart people coming out of Stanford don’t want to spend four-and-a-half hours copy-pasting between CMS systems. Now they get to spend their time understanding customers better, developing stronger storytelling and creating more differentiated messaging for different audiences.
AI is making it much easier to produce content at scale. Are we moving from a volume problem to an ideas problem?
One of the biggest conversations I’ve had with CMOs during Cannes is whether this is going to make the sea of sameness even worse or make us better storytellers.
I think it’s entirely true that AI — and agents particularly — have the capacity to create massive amounts of AI slop. I think we’ll probably see it get worse before it gets better. AI amplifies whatever core idea you have. If you have a brilliant idea and a beautiful story, it will amplify that. If you don’t, it’ll amplify that too.
The world is short on great ideas and beautiful stories, not on content. Our point of view is that agents allow us to become more human. They let us spend more time on the craft of storytelling, on understanding customers and connecting with people’s emotions.
Some marketers worry audiences can immediately tell when something was written by AI.
They can. Marketing has always been about having a clear brand strategy and a unique position in the world. Everything comes downstream from that. AI is just another variation of that. If you don’t have that core idea, this is garbage. You can simply produce more, faster, cheaper garbage.
What’s your advice to marketing leaders who are still trying to figure out AI?
You have to be hands-on. I create my own agents. I have my own collection of agents that help me get through my day. If you’re a CMO or any marketing leader who’s not doing that, you’re not going to be able to lead your team well. Once you engage with the technology, you realize how it can change the way you work for the better, not for the worse.
I even had my own moment. I have a positioning and messaging framework I’ve spent 15 years refining. We tried to build an agent around it, and it didn’t work very well. The team eventually came back and said, “The thing you love didn’t really work with the agent, so we built something different.” For a moment I thought, “All my skills are useless. I’m no longer useful.” But then you realize, no, you’re just going to work differently. Every marketer is going to have to change. But that doesn’t mean there aren’t wonderful new opportunities to create.
Is there one AI project inside AWS marketing that best illustrates this transformation?
One of my favorites is what we call Agentic Insights and Research. About a year ago, our data science organization had 1,500 custom dashboards. A new tool now allows any marketer to simply type something like: “I ran an AI campaign in France. How did it do?” The tool doesn’t just tell you the results, it gives you context, benchmarks, recommendations and explains what you should do next. It completely flipped the problem on its head.
What conversations around AI aren’t making it onto conference stages like Cannes?
One topic that’s being talked about more in technology circles than in marketing is sovereign AI. Particularly in Europe, customers are thinking about sovereignty, regulation and where their data resides. That’s why AWS has a European Sovereign Cloud that’s completely contained within Germany and managed by EU citizens. It’s incredibly important for regulated industries and government customers, even if it isn’t yet top of mind for most marketers.
If we’re sitting here again at Cannes in 2028, what will feel obvious then that still feels radical today?
It’s almost impossible to predict because the pace of innovation is so fast. Historically, AWS planned products on an annual cycle. Now it’s quarterly, and sometimes even faster.
A great example: We were planning an event about eight weeks out, and during that planning process, three entirely new products were built and launched at that event. We’re now at the point where it’s faster to build new products than it is to plan an event. That’s how quickly things are changing.
Finally, what’s next for AWS marketing?
One of the big things you’ll start seeing later this year is a refreshed brand strategy, along with a new visual language and design system. That will begin rolling out across all of our campaigns.