What should a CMO do first about AI?
Claudia Calori with Flavia Barbat and Thomas Marzano. Legible and Lovable, episode 7. Published by Brandingmag, 2026-10-28. 80 minutes.
Episode notes by Thomas Marzano.
Look past the vendors to your customers’ journeys, try the tools as a person rather than an executive, and train yourself before you brief your team. That is Claudia Calori’s advice to CMOs. She leads global marketing for Personal Health at Philips, and has spent the past year putting the Legible-Lovable Law to work there.
Start with the customer’s journey, and your own
Claudia and I were both at Cannes Lions this summer. In the side conversations I had with CMOs and brand leaders, everyone was taking steps, and the mood underneath was overwhelm: so much changing at once, a supplier or an agency for every piece of it. Claudia recognised the mood: leaders feel unequipped for the change and pressed to deliver the multiplier their boards were promised, without knowing where to start. She peer-reviewed the first draft of the manifesto Flavia and I published, so I asked her what she would tell a CMO in that position.
Her diagnosis comes before the advice. From the launch of ChatGPT in 2022 through 2025, the AI conversation in marketing was about the marketer’s journey: efficiency, faster and cheaper work, more content for more platforms. Claudia’s point is that marketers are social scientists, and the consumer’s journey is what should obsess them. The job itself hasn’t moved. In her words, a marketer creates meaning that creates preference, that triggers choice. What AI changes is where that choice is made, more and more often inside a conversation with an assistant.
She took the Legible-Lovable Law into her team as a mantra, and a few months ago gave it a working twist of her own: loved by humans, preferred by machines. Being read, she realised, is table stakes. The work is to make the shortlist an assistant draws up, and to be the option it recommends for the intent in front of it. Flavia asked whether that preference lasts. It doesn’t compound, Claudia answered; it is won case by case, every time someone asks. The Law reads as it always has: brands must be legible to machines and lovable to people. What has changed is the machine’s part in it. It returns a recommendation, and personal agents now infer what each of us would love, so the machine weighs both halves when it picks.
The clearest part of the conversation was a shaver. Claudia has yet to meet a customer who asks how many rotations per millisecond the blades perform. People want to feel confident before a wedding, a first day at work or a date. That story stays human. Underneath it, the product needs a structured, comparable layer that lets a machine match each feature to what the person is looking for, and lets the brand claim superiority, with integrity, only where it owns it. Then the brand stops talking about itself in the first person. The assistant says it instead: they are the best for you, for this reason. Flavia’s version is that the brand should be the one that teaches the machine what its claim means.
Trust is being pulled two ways at once. On one side, creators and the communities close to people decide more of what gets discovered and tried. On the other, machines leave no digital stone unturned, checking whether a brand has walked its talk, years back. The brand sits in between. My reading is that the two forces belong together. For thirty years digital interfaces took our time, so that is where brands learned to appear. As agents give that time back, brands have to matter again where they always should have: in the real world, in real experiences.
Inside a company, that scrutiny lands on functions far from marketing. A brand’s own voice now counts only so far, Claudia told her teams. It needs others to corroborate it: in her categories, clinical experts, dentists and the users who write reviews. If a shaver claims to shave closer and doesn’t, that is a quality problem and a marketing problem at once. Doing what you say, saying what you do and making sure others say it too is becoming an enterprise mandate. In the manifesto we called the failure the integrity gap: the costume brand that claims one thing and proves another. An agent finds that gap quickly, and it bites.
Then there is leading a team through a technology that isn’t finished. Claudia sorts AI work by three lenses: efficiency, augmentation, and what she calls the superhuman factor. Most of the value, she argues, sits beyond efficiency. In the least charted areas, such as GEO and agentic commerce, she treats the value as learning, captured by the team as it goes. The anchor is knowing what you stand for. Hers is helping people with their personal care routines, and a leader who is clear on that, and on where she won’t compromise, can sit down with an AI platform or a retailer already knowing what she wants from it.
Her three things follow from all of this.
- First, look past the vendors who each offer to fix one piece of your remit, GEO here, content creation there, and study how the journeys in your category are changing: are people starting their search in an AI assistant, and what are they asking it?
- Second, experience it yourself. Take the corporate hat off and notice which tasks in your own life, and in your friends’ lives, AI is changing.
- Third, be curious, and train yourself first. The old order trained the people in the weeds and briefed the top. With AI it runs the other way: put yourself through the training, compare notes with peers, and let your team see you trying things.
As Claudia put it, do it as if your job depended on it, because it does.
Questions this conversation answers
What should a CMO do first about AI?
Three things, in Claudia Calori’s experience as head of global marketing for Personal Health at Philips. First, be wary of the vendors who each offer to fix one piece of your remit, GEO here, content creation there, and look at how the journeys in your category are changing: are people starting their search in an AI assistant, and what do they ask it? Second, experience it yourself, as a person rather than an executive, and notice which tasks in your own life and your friends’ lives AI is changing. Third, train yourself first. The old order trained the people in the weeds and briefed the top; with AI it runs the other way. Put yourself through it, compare notes with peers, and let your team see you trying things. These are the leader’s own first moves. The brand’s first steps are in our FAQ.
Is it enough for AI to read your brand, or does it have to prefer it?
Being read is the start. Claudia’s team works to a version of the Law she adapted for them: loved by humans, preferred by machines. Being read by machines, she realised, is table stakes. What counts is making the shortlist an AI assistant draws up, and being the option it recommends for the intent in front of it. That changes the work. Being read can be handled as due diligence, with content kept in order; being preferred asks for a structure that shows why you solve this person’s need better than the alternatives. The Legible-Lovable Law itself reads: brands must be legible to machines and lovable to people. What the assistant returns is a recommendation, and personal agents now infer what each of us would love, so the machine weighs both halves when it picks.
Can a brand stay preferred by AI, or does it have to win every time?
It has to win every time. Flavia raised the doubt: legibility felt like something a brand controls and can plan for, while preference, judging by how AI visibility is measured, doesn’t come back the same on every query. Claudia gives preference no compounding effect. A brand doesn’t become the machine’s favourite and get recommended back on repeat; it wins instance by instance, as the best solution for the specific intent. The battle, in her words, is never won once and forever. For her team that moves the work from content hygiene to structure: the brand is codified so that its case can be made each time a need emerges. Why recommendations vary from one asking to the next is the subject of Episode 4.
Listen from 16:35 · Episode 4: why AI recommends some brands
What does AI need to know about a product that customers never ask?
The specifications your customers skip. Take a shaver. Its product pages are packed with rotations, water resistance and other specifications, and Claudia has yet to meet a customer who asks how many rotations per millisecond the blades perform. People care whether they will feel confident before a wedding, a first day at work or a date. An AI assistant cares about both, because it compares your specifications against your competitors’ on the person’s behalf. So the human story stays on top, and underneath sits a structured, comparable layer that maps each feature to the needs it serves, such as sensitive skin or a special event, and claims superiority only where the brand owns it. Then the agent, not the brand, says it: they are the best for you, for this reason.
What is a marketer’s job when AI mediates the choice?
The same job, aimed at the right journey. Claudia’s definition: a marketer creates meaning that creates preference, that triggers choice. Content is a means. Marketing is close to a social science: you study how people live, find the tensions and problems a product or service can solve, and earn trust by solving them reliably, again and again. Since ChatGPT launched in 2022, most of the AI conversation in marketing has been about the marketer’s own journey: efficiency, speed, more content for more platforms. The bigger change is on the consumer’s side, where AI is reshaping how people discover and choose. If marketing is about solving people’s problems reliably, the machine’s role is to surface the brands that do it, consistently, for the need in front of it.
Who do customers trust when creators and AI both shape what they buy?
Less and less the brand, which now sits between two forces. On one side, creators, influencers and local communities decide more of what people discover and try. Claudia pointed to the pull of proximity, such as the REWE store in Wiesbaden that grows basil in a greenhouse on its own roof and sells it in the aisles below. On the other side, AI parses every digital trace to check whether a brand has walked its talk, and an old misstep can resurface at the moment it matters most. Both answer the same question: whom do I believe? My reading is that the forces are intertwined. Digital interfaces took our time for thirty years, so brands learned to appear there. As agents give that time back, brands have to be relevant where they always should have been: in the real world.
Why is product quality now a marketing problem?
Because the proof now travels through other people’s voices, and AI reads them. Claudia has told her teams that a brand’s own voice matters only up to a point, for people and machines alike. Brand building now means having that voice corroborated by others: for a personal health business, clinical experts, dentists, and the users who write reviews and complaints. So quality, traditionally far from marketing, is where the promise is kept or broken. If a shaver claims to shave closer than any other and doesn’t, it is a quality problem and a marketing problem at once. Doing what you say, saying what you do, and making sure others say it too is becoming an enterprise mandate: the integrity gap, managed across the company.
How should a CMO show the value of AI when the technology isn’t finished?
By separating the kinds of value, and by being open about which are still being learned. Claudia sorts her team’s AI work by three lenses: efficiency, augmentation, and what she calls the superhuman factor. Efficiency is where most enterprises start, in content production, consumer care and insight generation. The larger value, she argues, sits in augmentation and beyond, where it meets how enterprises work: handovers and structures that predate AI. In the least charted areas, such as GEO and agentic commerce, she treats the value as learning: experiment with courage, capture what the team learns together, and expect some assumptions to fail. What keeps that from drifting is a clear view of what you stand for. A leader who knows where she won’t compromise can negotiate with an AI platform or a retailer already knowing what she wants.
Why does brand consistency matter more once AI is reading?
Because consistency is what both people and machines remember. In the Law, lovable is not the brand love of the marketing literature, the fandom and retention of another era. It is whether a person relates to the brand, and whether it fits their problem and their taste when they choose. Claudia makes it actionable in a similar way. Love, for her, works as a mental shortcut: the memory structures that bring a brand to mind first at the moment of choice. An AI builds its picture of a brand from the residue that brand has left across many sources, so the same consistency shapes how it renders the brand. Her hygiene rule is consistency of message, of key claims and of point of difference. The wording can change. You can refresh the chapters, as long as the story stays the same.
Calls made on air
Published 28 October 2026; windows added in November 2026. Each call as I made it, with the window in which I expect it to land: the year it could first happen, and the year by which I think it will have happened.
- Open models will run locally on our laptops and phones, as agents that sit on our own devices rather than services such as Muse and dots. (55:19) From 2026, by 2027.
- Personal agents infer what each of us would love, so the machine now carries both halves of the Legible-Lovable Law. (41:24) Holds now.
- As agents give us back the time screens took, brands must be relevant again in the real world. (35:22) Direction of travel.
In their words
“What you want to do as a marketeer is create meaning that creates preference, that triggers choice.”
“Eventually, what I realized is that read by machines is actually table stakes.”
“That battle is not won once and forever.”
“It will not be the brand to talk about itself in first person. It will be the agent to say, ‘Hey, they are the best for you for this and this reason.’”
“I want to be the one that teaches the machine what that means.”
“Brands need to be relevant yet again, where they always should have been relevant, which is in the real world.”
Chapters
- 00:00 Meet Claudia Calori
- 07:32 What is a marketer’s job when AI mediates the choice?
- 14:19 Loved by humans, preferred by machines
- 16:35 Can a brand stay preferred by machines?
- 21:02 What does a machine need to know about a shaver?
- 28:43 Proof, creators, and the two forces shaping trust
- 34:26 Is lovability the same as brand love?
- 40:35 What do personal agents like Muse and dots change?
- 49:07 Who is mediating for whom?
- 55:46 How do you lead change when the technology isn’t mature?
- 1:05:27 Is brand governance becoming everyone’s job?
- 1:10:04 Where should leaders start? Three things to focus on
Sources
- REWE Group, “REWE Green Farming” (press release), 27 May 2021
- Brandingmag, “Brand Constitutions: The Legible-Lovable Standard for Building Equity in an Agentic Economy”, November 2025
Where to start
A CMO’s first moves are personal; the brand’s come next. Seeing how AI assistants describe your brand today, and whether other voices corroborate what you claim, is where the Agentic Brand Readiness Assessment begins.
Explore the engagements · Start a conversation
About the series
Legible and Lovable is Brandingmag’s podcast on agentic branding, hosted by Flavia Barbat, editor-in-chief of Brandingmag, and Thomas Marzano. In Brandingmag’s words: “Thomas identified the field’s dynamics and wrote the initial thesis; Flavia named the field and edited the content.” Episode 7 is the series’ first with a guest. Claudia Calori is vice president and head of global marketing for Personal Health at Philips, and has built brands at Procter & Gamble, Heineken and Converse. She peer-reviewed the first draft of the Brand Constitutions manifesto. Thomas and Claudia worked together at Philips, where he was global head of brand.
Where this sits. Agentic Branding is the discipline of making brands intelligible, governable, and desirable in an AI-mediated world. Its governing principle is Thomas Marzano's Legible-Lovable Law: brands must be legible to machines and lovable to people, with the machine now carrying both halves.
All episodes of Legible and Lovable. Episode notes by Thomas Marzano, updated 2026-11-03.