The Absence and the Manner

L'Oréal × Kering, March 31. 4 billion euros for Creed and Kering's beauty business, with a longevity joint venture written into the agreement. Luxury formally states that the body is a vertical.

Zoï. The Parisian longevity clinic (Niel, Ducasse, Saadé, Moueix, Bancel among its backers) hands its medical governance to Professor Jérôme Salomon, France's former Director General of Health. The agent measures, the proper name answers.

Oura, May 21. Confidential IPO filing, 11 billion dollars in valuation, 5.5 million rings sold. Self-measurement becomes a listed asset.

My footprint audits, two verticals. A Parisian fashion house and a Luxury hospitality network, cited when you name them, found one time in two when the client does not. Discovery is the lost territory.

EU AI Act, May 7. Digital Omnibus agreement: transparency held at August 2, 2026, obligations on high-risk systems pushed back to December 2, 2027. The calendar slips, the exposure stays (15 million euros or 3% of worldwide revenue).

The more the agent learns to measure, compare, recommend, the less what it measures is worth. Your watch runs an electrocardiogram, a Swedish clinic sells a full-body scan for a few hundred pounds, any phone reads a night's sleep. Data becomes commonplace at the speed it democratizes.

And as it becomes commonplace, value shifts elsewhere. Toward what no agent does alone. Putting a name on the line. Translating a number into a decision within a singular life. Looking at someone. Luxury picks the wrong merchandise if it sells measurement: its rent is presence. But you still have to exist in the machine's answer before you can embody it in the store. The first battle is not the sale. It is the answer.

Sources:

• Neko Health: full-body scan from £299 (Stockholm, London).
• Apple Watch: built-in electrocardiogram and health sensors (FDA clearance since 2018).

L'Oréal buys Kering's beauty business, and bets on longevity

On March 31, 2026, L'Oréal finalized the acquisition of Kering Beauté for roughly 4 billion euros: the perfume house Creed, and fifty-year licenses to develop the beauty lines of Bottega Veneta and Balenciaga (Gucci will follow once the contract with Coty expires, expected around 2028). A detail no one comments on, and which should keep you awake: the agreement, sealed back in October 2025, includes a joint venture explicitly dedicated to wellness and longevity. The two biggest names in beauty and in Luxury are not merely buying brands. They are naming a frontier. For prestige beauty, the consequence is clear: the skin diagnosis must become an agentic follow-up over time, not yet another app forgotten after three uses.

Sources:

• L'Oréal Finance, press release, 10/19/2025 and 03/31/2026
• Business of Fashion, 04/01/2026

Oura files for its initial public offering

On May 21, 2026, Finland's Oura filed confidentially for its planned IPO with the SEC, the US securities regulator, on a valuation of about 11 billion dollars (reached late 2025 in a 900-million round), 5.5 million rings sold, and revenue its CEO projects near 2 billion dollars in 2026 (the official target hovering around 1.5 billion). The body becomes a financial asset. But a valuation measures investor appetite, not the effectiveness of an agent. What you should take away is the brand lesson: co-signing a third-party sensor, as Gucci did with the Oura ring, brings prestige, and lets a partner read your client's most intimate data, her sleep, her heart, her stress. The prestige is yours. The data is his.

Sources:

• CNBC, 05/21/2026
• Bloomberg, 05/21/2026

WHOOP: you no longer buy an object, you subscribe to an agent

Valued at 10.1 billion dollars at its March 2026 Series G, WHOOP does not sell a band, it sells a follow-up: the sensor has no screen at all, you pay for the subscription. Its agent, WHOOP Coach, powered by OpenAI's AI, does not merely display data. If your night was short, it replaces the strength session with recovery on its own. It decides. The lesson is not the sensor, it is the shift: from the one-off sale to the continuous relationship, held by an agent. The question for your House is the twin one: do your agents hold the relationship as well as WHOOP holds its client's body?

Sources:

• Bloomberg, 03/31/2026 (WHOOP Series G at 10.1 billion dollars)
• Subscriber figures disclosed by WHOOP

One word, which will return in every edition where the client's discovery plays out at an agent rather than in your shop window. Not a refusal of AI. What you win, or give up, when a client asks a machine for advice.

Keeping your clients, not only winning them

Look at the two models that Zoï and WHOOP embody, because they herald yours. WHOOP turned the purchase into a subscription: value is no longer in the object, it is in the follow-up that lasts. Zoï, a stone's throw from Place Vendôme, charges 3,600 euros for a 145-biomarker assessment, and its most strategic move is not technological: the clinic handed its medical governance to Professor Jérôme Salomon, France's former Director General of Health. A clinic at that price is not buying one more clinical skill. It is buying a signature that answers. There is the winning pair you will see everywhere: the agent measures, the human puts their name on the line.

Why this concerns you, figures in hand: Luxury sales concentrate on the minority of best clients, to the point where the top 0.1% alone accounts for 37% of the value (Bain-Altagamma, 2025), and at Mytheresa, 39% of the business volume comes from 3.5% of its clients (Top Customer program, fiscal year 2023). Your margin lives off a handful of clients you have to keep, not win, even as Luxury's client base contracts. And "keeping" a client over time is precisely what health has just re-industrialized. But before you can keep, you have to be found.

Before keeping, you have to be in the answer

AI does not kill your sale, it shifts the first move. According to McKinsey, the client gladly delegates discovery to the agent (finding, comparing), but not the purchase of an object of desire, where human presence is part of the value. The implication is plain: the agent presents a first selection, the human closes the sale. If you are not in that selection, no one closes. And let no one on your committee tell you the battle is already decided: the agentic transaction remains embryonic, OpenAI quietly dropped its Instant Checkout in March 2026, barely 8% of users had tried it, for want of buyers. But discovery is already in play. That is the front to hold today.

The client trusts the agent to find. They trust the human to choose.

— My study The client, the brand, and the agent, April 2026.

This is the terrain of GEO (Generative Engine Optimization, the art of being correctly captured by the agents that speak in your place). It is not search ranking: an agent does not present a list of links, it takes the floor. Everything that has not been said about you in the corpus that feeds it does not exist in its answer. Everything that has been said wrong passes for true. And beware, being cited is not enough. Cited on price, you become commonplace; presented as a desirable world, in your own words, you protect your brand. The footprint is measured in desirability captured, not only in queries won. And its raw material is not your website: it is the press. Close to 85% of what an agent cites comes from earned media, from third-party sources, not from the pages you control (Muck Rack, 2026). Your narrative is won first where others speak about you, and that is precisely the terrain of your communications department.

Cited on price

"For a well-made leather bag, expect 2,000 to 4,000 €. [Your House], [Competitor A] and [Competitor B] offer comparable models: the choice mostly comes down to your budget."

A line in a comparison engine. Interchangeable.

Cited as a world

"[Your House] has carried on a rare saddlery craft since 1837: each piece is hand-stitched, with the saddle stitch, and gets passed down. It is not a bag, it is an heirloom object."

Desired, in your own words. Preferred.

The two guardrails: own your agent, secure your data

First rule: own your agent, or a third party will speak about you without you. The alternative to the co-signed sensor exists: the brand-authored agent, written and hosted by the House, like Ask Ralph at Ralph Lauren (edition #1), which keeps both the narrative and the data. Your school of thought changes the dose, not the necessity: Hermès lines up record after record while keeping AI discreet, everything indicates it stays confined to logistics and anti-counterfeiting, far from the sale; LVMH industrializes more openly, while drawing the line: when Google launched its agentic cart Universal Cart at Google I/O 2026, Sephora, its beauty retail banner, was among the first connected merchants, but not a single object-of-desire House from the group appeared in the list initially published, neither Vuitton nor Dior. Both are right on their own terrain. Neither can afford to be absent when a client asks an agent.

Second guardrail: where the data lives, and who signs. As soon as an agent touches a health data point, an intimate orientation or a biometric data point, you fall under Article 9 of the GDPR: legal basis, explicit consent, impact assessment; the other client data that is sensitive in a commercial sense, measurements, wealth, habits, stays under the general regime, already demanding. And if it decides alone, with no real human intervention and with a significant effect on the person, that is Article 22. The EU AI Act keeps the transparency obligation at August 2, 2026; the Digital Omnibus agreement of May 7, 2026 pushes the obligations on high-risk systems back to December 2, 2027 (subject to formal adoption), but the exposure remains: 15 million euros or 3% of worldwide revenue. And the data is hosted somewhere: WHOOP runs on OpenAI, Amazon One Medical on Bedrock, hence within reach of the US CLOUD Act. For a House, the sovereignty of client data is not a legal detail, it is a selling argument. Besides, when the free agent gets it wrong, and it does (an independent assessment by Mount Sinai published in Nature Medicine finds that a consumer health assistant under-triages one emergency in two), value shifts toward the human-supervised agent. True in health. Soon true for advice from your House. The real question no one asks on the committee: who signs when the agent gets it wrong about your best client? And the day the agent no longer prepares the answer to the client but evaluates your own advisors, you tip into the "high risk" of the EU AI Act, mandatory human oversight: training the advisor to work with the agent is one thing, governing the agent that scores the advisor is another, sensitive in a quite different way.

What no agent does alone

Veblen described conspicuous consumption in 1899. A new grammar is added to it, conspicuous health: you no longer flaunt what you buy, you flaunt what you inflict on yourself, the fast, the ice bath, the four-figure assessment. And the grammar is not universal: the Asian client readily displays their health score where the European cultivates it in secret, an agent blind to that nuance makes you miss, precisely, the client who matters most. But measuring becomes commonplace. What stays rare, and therefore what you must sell, are the three things no agent does alone. Responsibility: a proper name that puts its reputation on the line behind the decision, and remains reachable when the agent gets it wrong (what Salomon brings to Zoï). Meaning: translating raw data into a decision, within a singular life, because data creates anxiety and meaning sets free. Time: the attention of a human who looks at you.

What needs to happen now

Four moves, in order, for a CEO and a digital director.

1. Measure your footprint across the six intents of the journey (discovery, comparison, knowledge, purchase, ethics, retention). You do not fix what you have not measured, and the diagnosis costs little where traction has not yet taken off.
2. Write your agentic narrative and own the agent that carries it: what the machine is allowed to say about you, in your words.
3. Draw your Delegation Grid, validated by leadership, and break it down by function.
4. Appoint someone in charge on the executive committee, not a data protection officer tucked away in the third basement, but a member of the ExCom who answers, to the CEO, for the arbitrations between agentic AI and what is never delegated.

Sources:

• Bain & Company × Altagamma, Luxury Goods Worldwide Market Study, 2025: concentration on the best clients, contraction of the base.
• Mytheresa (LuxExperience), FY2023 results: 39% of GMV generated by 3.5% of clients (Top Customer program).
• Le Quotidien du Médecin: Prof. Jérôme Salomon joins Zoï to consolidate its medical governance, January 2026. The 20 M€ funding round (Niel, Ducasse, Saadé, Moueix, Bancel) documented by Maddyness and Stratégies.
• McKinsey × Business of Fashion, State of Fashion 2026; McKinsey, When AI Meets Desire.
• Mount Sinai (Icahn School of Medicine), Nature Medicine, 02/23/2026: ChatGPT Health under-triages 52% of emergencies (NBC News summary).
• Google, launch of Universal Cart, Google I/O, May 2026.
• EU AI Act, agreement of May 7, 2026 (Hogan Lovells).

This spring, I put two Houses to the test of the agents, in two worlds that nothing brings together: a contemporary Parisian fashion house, a runway label, and a large international Luxury hospitality network. Same conditions each time: five agents in their consumer models (ChatGPT, Claude, Gemini, Perplexity, Mistral), 100 calibrated questions (50 in French, 50 in American English), six client intents, six weighted criteria, NarrativeGuard method. Five hundred real answers per House. Both audits are confidential: I cite them anonymized, the names change, the findings do not.

Two worlds, the same blind spot. Both Houses score an overall agentic desirability of around the average, grade C. A word on that grade: the audit scores each House out of 100, then translates the result into a letter, from A to F like a school report card. A, the House that agents render faithfully and recommend; F, the one they ignore or distort. C is the middle: the House exists in the machine's answer, but without imposing itself there. And for both, the score always breaks at the same place. When the client names the House, the agents know almost everything: it is cited in nearly all answers, its history and its codes rendered accurately. But when the client only describes what they are looking for, without naming the House, the one who is discovering, the one who will make up your clientele of tomorrow, it emerges only one time in two.

You are a house people find again, not a house people find.

Why this should keep you awake: on the perimeter I audited in contemporary ready-to-wear, on the order of 70% of conversion comes from this non-intentional discovery, from the client who was not looking for you. This is an observation drawn from my audits, not a market average, but it conveys the scale of the stakes. It is exactly the half of the conversation you are not holding. And when you do not emerge, the space does not stay empty: it is occupied by a competitor, or by a third-party reference that speaks in your place and reaps the credit you should be carrying.

A more reassuring finding: the gap in treatment between French and American English is small, two points out of a hundred. For Houses that generate a large share of their revenue outside France, that is a rare asset. The work, then, is not linguistic. It is one of presence and narrative. These Houses do not have an awareness problem, their knowledge is near perfect. They have a problem of discovery and of defending their narrative. Exactly the zone the agent governs, and that most Houses do not yet measure.

Method & sources:

• Proprietary footprint audits, confidential and cited anonymized, NarrativeGuard protocol: 5 agents (ChatGPT, Claude, Gemini, Perplexity, Mistral), 100 calibrated questions in French and American English, 6 intents and 6 weighted criteria, 500 answers scored per House, spring 2026.
• Study The client, the brand, and the agent (April 2026, 680 answers): public methodological foundation.

This is precisely what the Flash Agentic Footprint Audit measures. And it does not stop at your share of voice, at the single question "am I cited?" It measures your desirability as the agents render it, on a grid of six weighted criteria: presence, position, accuracy, narrative preservation, actionability and sentiment. The setup: five agents (ChatGPT, Claude, Gemini, Perplexity, Mistral), 100 calibrated questions in French and American English, 500 answers scored per House across the six intents of the client journey, against your real competitors. The outcome: a recommendation and an operationalized, sequenced action plan, delivered to your executive committee in one to two weeks. Built to decide, not to decorate.

Measure my footprint →

And if the audit reveals a narrative project, we draw your AI doctrine and your Delegation Grid together. A single interlocutor, from measurement to decision.

To the marketing director. When an agent recommends a House like ours, do we come out, and in our own words, or as substitutes that level us down?

To the client director. Across our 100 best clients, does the agent keep the relationship from one time to the next, or convert it only once?

To legal and IT. On which data, hosted where, and who signs the day the agent gets it wrong about your best client?

To the head of HR. Do our advisors know how to work with an agent that prepares their recommendations, and do we know how to take a pilot to scale?

Not asking these questions does not erase their answers. They are being written already, at this very moment, in what the five agents say about you while you read these lines.

• McKinsey × Business of Fashion, State of Fashion 2026. The wellness figures at the source, and the gap between discovery and purchase in Luxury.
• McKinsey, When AI Meets Desire. Why the client delegates the search to the agent but keeps the desire purchase for the human.
• Bain & Company, Luxury Goods Worldwide Market Study (2025). The concentration on the best clients and the contraction of the base.
• Kering and L'Oréal, the beauty and longevity alliance. The 4-billion agreement and its wellness joint venture, at the source.
• The Digital Omnibus and the EU AI Act, agreement of May 7, 2026. For legal and IT: what slips, what stays.
• My study The client, the brand, and the agent (April 2026). 680 answers, the new triangle of client, brand, agent, and the boundary between finding and choosing.

Next week, I don't yet know what the news will put on the table. I only know where I will be looking: at the exact spot where an agent decides, in your place, what a client will see of your House.

Until then, run the test. Read what the machine says about you. And ask yourself who, behind the answer, will put their name on the line. A House worthy of the name owes it to itself to be in the answer, and to answer for it. Luxe oblige!