Before the Storefront

Before the Storefront

Stendhal described it in 1822 in On Love: crystallization. A dead branch plunged into the salt mines of Salzburg emerges covered in crystals. Every imperfection in the wood is coated. This is how desire forms around an object: not in its presence, but in its absence. In the waiting. In the imagination of the not-yet-possessor.

Luxury houses are machines built to manufacture that waiting. The campaign that makes you dream without showing the price. The window display that stops you without forcing entry. The advisor who slows the transaction to intensify the desire. Every obstacle is an additional layer of crystals. Friction is not a flaw in the journey. It is the product.

There now exists a space between the client's intent and the house's storefront. A space the houses did not build, do not control, and where their future clients are spending more and more time. Call it the Antechamber.

Where desire now forms

In the Antechamber, preferences form before the visit. Comparisons happen without the brand. Decisions are made — or nearly — before the house has had a chance to seduce. The numbers give the measure: 84 million shopping queries per week on ChatGPT in the United States (Modern Retail / Stackline, March 2026). 700 million weekly active users worldwide (eMarketer). Two-thirds of Gen Z buyers now use an LLM (Large Language Model) to research products before purchasing (Harvard Business Review, 2026). And this data is often inaccurate — which makes the problem worse: the house is not merely absent from the Antechamber, it is potentially misrepresented there.

Jing Daily documented it on March 20: "AI-powered systems are making purchasing decisions before consumers ever browse — and brands that fail to adapt risk being misread, not just overlooked." The risk is not invisibility. It is distortion.

McKinsey, in "Europe's Agentic Commerce Moment" (March 2026), details the mechanics: discovery is now the phase where preferences form and winners emerge. Agent usage declines as you approach the transaction — the final purchase remains human, the desire no longer entirely so. The decisive moment has shifted upstream. And OpenAI confirmed it at its own expense on March 24 by withdrawing Instant Checkout: the locus of power is not the register. It is the recommendation that precedes it.

Baudrillard anticipated it in Simulacra and Simulation (1981): the sign can precede and replace the reality it was supposed to represent. This is exactly what is happening in the Antechamber. The AI agent's response is not a path to the house — it is, for that client at that moment, the house itself. A house whose words it did not choose.

Steve Jobs put it differently: "People don't know what they want until you show it to them. (...) Our task is to read things that are not yet on the page." Luxury houses have been doing this for centuries — reading what is not yet written, anticipating desire. The difference: they did it in the storefront, in the salesperson's gesture, in the silence of the showroom. The AI agent does it in the Antechamber — and it does it before them.

This is not a crisis of desirability. It is a crisis of the geography of desire.

Sources: Jing Daily, March 20, 2026; McKinsey "Europe's Agentic Commerce Moment", March 2026; Modern Retail / Stackline, March 2026; Retail Gazette (OpenAI pivot), March 24, 2026

Being cited is no longer enough

Converteo measured ten houses on ChatGPT and Gemini in March 2026 — accounting for 86% of the global AI audience — across five dimensions: visibility, narrative fidelity, factual accuracy, actionability, source quality (GEO Benchmark 2026, full downloadable report). Result: luxury house visibility in LLMs is still heavily tied to their financial valuation, their heritage, and their media capital — rather than the relevance of their current offerings.

Actionability — the ability of AI responses to convert into a visit or a purchase — is the lowest score for every house (39-52/100). The best-represented owe it to 150 years of writings, reviews, and digitized archives feeding the language models. This is not a strategy. It is a rent.

Bourdieu would say the great houses were living off symbolic capital accumulated over a century and a half. This capital precedes them — it belongs to them no more than an inheritance belongs to the one who receives it. And capital that is not actively maintained — in the vocabulary of the models, in the quality of agent-readable data, in actionability — erodes. Silently.

The data confirms this from other angles. Avenue Z and Profound now measure "Citation Share" — a brand's share of voice in LLM responses — and find that roughly 10% of referral traffic for fashion brands already comes from conversational AI engines (Avenue Z / Profound, March 2026). Relevance Digital notes that UHNW (Ultra High Net Worth) consumers adopt AI search at twice the average rate (Relevance Digital, March 2026). Gartner forecasts a -25% drop in traditional search engine query volume by the end of 2026. Traffic is migrating. And the houses that have not built a presence in the Antechamber will not even see it leave.

Being cited in the Antechamber is not enough. You must be actionable there. And on that front, no one has an answer yet.

Sources: Converteo "Luxury in the Age of AI Search 2026"; Avenue Z / Profound, March 2026; Relevance Digital, March 2026; Harvard Business Review, 2026

Two doors, one hallway

Sephora enters ChatGPT through beauty — LVMH's most data-rich vertical, the least risky to test. Kering appoints a CDO whose roadmap explicitly mentions agents. Two groups, two moves, same reading of the problem: the Antechamber exists, and you need to be in it.

The infrastructure follows. Gap installs checkout inside Google Gemini via the Universal Commerce Protocol, without redirect — Google Pay, AI sizing by Bold Metrics. Sven Gerjets, CTO of Gap: "It's no longer the keyword, it's the conversation." (CNBC, March 24). Accenture invests in DaVinci Commerce and introduces the concept of "Agentic BrandStore": conversational brand experiences governed inside the LLMs themselves (Accenture, March 23). At DFS (LVMH travel retail), Rezolve AI reports +10% conversion and +50% search revenue via its agentic engine (Globe Newswire, March 19). Soumia Hadjali, at Louis Vuitton, sums it up: "Agentic commerce is about being where [customers] are and being relevant at the right moment."

What distinguishes the two approaches at the top: Sephora enters the external agent. Kering, via Balenciaga × Intelo.ai, builds internal agents to optimize merchandising. Neither answer is yet that of the heart of luxury — jewelry, high-end fashion, leather goods. Deloitte Digital quantifies it (March 2026): roughly 40% of houses are still evaluating how to deploy AI, 41% have begun implementing it in selected areas, and fewer than 12% integrate it into their core functions. The "permanent pilot" remains the dominant mode in the sector.

These houses are watching. They are right to take their time. They are wrong to believe they have it.

Sources: CNBC (Gap × Gemini), March 24, 2026; Accenture, March 23, 2026; Rezolve AI / DFS, March 19, 2026; Deloitte Digital, March 2026; Retail Brew / Shoptalk, March 27, 2026

The asymmetry of attention

Everything above is infrastructure. What follows is about the bond.

Luxury rests on a double gap in attention. The first is an asymmetry: the house chooses who receives its gaze, the client does not — they aspire. The second is an asynchrony: desire is born in the client long before the house returns it. The client desires, waits, qualifies themselves; the house observes, evaluates, and responds only at the moment it has chosen. The waitlist is not an operational accident — it is the institutionalized form of this asynchrony. Stendhal called it crystallization: desire thickens in the temporal gap between the craving and the object. The houses that open the door to the AI agent introduce into this relationship an intermediary that abolishes both gaps at once. The agent is symmetric: it gives anyone the information the house reserved for the client who had crossed the threshold. And it is synchronous: it answers in three seconds. The time that manufactured desire — the time the house had instrumentalized for two centuries — disappears.

Capgemini, in "From Hype to How" (February 2026), poses a figure that should temper enthusiasm: only 8% of consumers would pay for an autonomous AI shopping agent. Meanwhile, 74% say they value human interaction in-store. The luxury client does not demand fluidity. They demand consideration. Daniel Langer puts it another way: "In a world where machine intelligence is abundant, how do you build a brand that is worth more because it is irreducibly human?"

Still, the projections are there. Bain, via The Fashion Law, estimates that 15 to 25% of US e-commerce sales will pass through AI agents by 2030 — that is $300 to $500 billion (The Fashion Law, March 2026). Morgan Stanley projects 50% of online purchases via agents at the same horizon. 30 to 45% of US consumers already use GenAI (generative AI) for product research. The trend is not reversing.

I am convinced that the houses that master the Antechamber will not be those that opened the widest door. They will be those that learned to control what happens in the hallway — by deciding which data the agent can read, which narratives it can tell, and which thresholds it may not cross. Friction has not disappeared. It has changed form. The question for every house is now: are you eliminating it, or redefining it?

Sources: Capgemini "From Hype to How", February 2026; The Fashion Law / Bain, March 2026; Morgan Stanley, agentic commerce projections 2030