Coca-Cola's Javier Meza and Maira Genovese on 'Test and Learn' AI Advertising

LISBON–Artificial intelligence has dominated the agenda at this year’s Web Summit in Lisbon. In one of the more grounded sessions, Coca-Cola’s European marketing chief, Javier Meza, and agency founder Maira Genovese used their fifteen minutes on stage to talk less about futuristic breakthroughs and more about the work of running large-scale marketing systems at speed.

The fireside chat, moderated by Steve Antoniewicz of trade publication The Drum, offered no dramatic revelations. It did, however, give a clear view of how a global consumer group and one of its agency partners are using AI to accelerate routine tasks while trying to protect creativity and trust.

Antoniewicz opened with a nod to current fashion. “In the future this might take place with some avatars, but today thankfully not,” he said, introducing “two very intelligent humans” to “talk about harnessing creativity before AI takes over”.

Asked whether machines will “take over” creativity, Meza was firm. “I would say no.” AI will continue to change “how we do things, especially in marketing and creativity”, he argued, but “it will not change why we do things”. The purpose, he said, remains “understanding humans, satisfying benefits, expectations, needs and aspirations”, and “creating relevance, excitement and inspiration”.

He was careful to distinguish process from ideas. “We use AI for content creation, which is not the same as creativity. I emphasise this: content creation and creativity are not exactly the same.” Coca-Cola’s applications, he said, are largely practical: “We are using AI for insight harvesting and predictive analytics to improve how we allocate resources. We also use it in supporting functions, from summarising meetings to creating presentations.”

So far, the gains are concentrated in efficiency. “I really believe we are still in the phase of faster and better, not yet cheaper,” he said. Longer-term savings, he suggested, depend on redesigning how marketing is organised. “The value capture of AI will take a couple of years, when we redesign structures, teams and partnerships with agencies. Today we are still doing interventions, picking parts of the process and using AI there.” He compared the shift to the early years of electrification. “Electricity took about 40 years from the first light bulb to capturing the full value. You had to redesign processes and organisations. It will be similar here.”

AI and the myth of perfection

Meza was candid about the uncertainty that comes with that experimentation. “Are you 100 per cent sure this will work?” he said his team sometimes ask. “The answer is, of course not. Sometimes you project confidence so you can push the boundaries.” The point, he argued, is to treat mistakes as part of the cost of learning. “On ethics and trust, transparency matters,” he said. “We recently put on air an AI-generated Christmas ad and we were very clear that it was AI-generated.” Viewers quickly spotted small inconsistencies. “Mistakes will happen. Fix them and move on. The technology is evolving very fast, and we are learning fast.”

Coca-Cola has chosen to move early rather than wait for perfect frameworks. “We decided not to spend two or three years only strategising,” Meza said. “We jumped in with partners, within ethical limits, willing to experiment, make mistakes and embrace the debate.”

He offered a glimpse of how other consumer groups are using similar tools. “A global brand told me they are using AI in the drive-through to understand the customer in real time and guide the conversation,” he said. “We are still in the business of selling oxytocin, which comes from human interaction. We are using technology to make that interaction even better.” For Meza, that remains the test: lighter workflows and richer human moments.

‍

Client expectations and category nuance

From the agency chair, Maira Genovese, founder and chief executive of MG Empower, kept her emphasis on client fit and responsibility. “As an agency our responsibility is to listen to clients. They are pushing us to add AI to how we create, but we must not just follow the trend,” she said. “If we add AI to the creative process or strategy, we need to do it with responsibility and without losing the value of the service we offer.”

Clients, she noted, often want the excitement of a new tool as well as the reassurance of familiar emotional cues. “Clients may want technology because it is exciting, but we come back to human connection and emotion. There has to be a balance between both.”

Genovese argued that sector differences still matter. “In luxury the motivations are different. Many luxury clients value craftsmanship and the emotional connection to the brand. They do not necessarily want technology at the point of purchase,” she said. Beauty is more experimental. “People want brand experiences. They want to go to events and play with the brand. They also want the brand to use AI inside the experience.”

Her own team, she added, is using AI mainly to support scale. “We use AI to help with process and scale, especially with content, without losing quality, local nuances or storytelling. It is test and learn, with risk assessment and privacy in place.”

An operating model

Many of the themes echoed comments Meza made in an earlier profile with THE GOODS, where he warned that “AI is going to continue disrupting our lives and our work” but argued that “fundamental human insight does not vanish just because the tools are different”. There too, he underlined the operational angle, describing how Coca-Cola used new production technology to remake its “Holidays Are Coming” Christmas campaign in a month and at significantly lower cost, freeing budget “to be used more strategically”.

In Lisbon, the emphasis was similar. The Web Summit session did not try to declare a creative future in a quarter of an hour. It did something more prosaic and, for many marketers, more immediately useful. It outlined an operating model: use AI to clear the undergrowth of process, redesign organisations slowly to capture the full value, stay transparent about errors and keep the emotional core of brands firmly in human hands.

Or, as Meza and Genovese both suggested in different ways, systemise the boring, humanise the meaningful, and hold the line on taste.