How five AI models perceive, surface, and narrate ten luxury fashion Maisons.
The March AI Visibility Index measured whether AI surfaces the great Maisons at all. The Fragrance edition looked at how it values them once they surface. This Fashion edition takes on a question neither had to face: when AI describes a Maison with confidence, is it describing the Maison that exists today. We read ten fashion Maisons on two axes. AI Presence is how reliably they surface. Cultural Desirability is the heritage, savoir-faire and cultural authority the models attribute to them. We then audit a third thing the desirability score leaves alone, whether the facts are current.
Across 2,400 responses on Claude, ChatGPT, Perplexity, Gemini and Google AI Overviews, the strongest pattern is timing. In 82 mentions, the models name an artistic director who has already left, and those errors land on the Maisons whose creative direction changed most recently and most publicly. AI is most fluent about the Maisons in transition, and most confidently wrong about them too.
The director-staleness audit is evidentiary and kept strictly separate from desirability scoring. Full methodology at the end.
The vertical divider is the sector median for AI Presence (50.2); the horizontal divider, the median for Cultural Desirability (67.0). Four zones emerge, and heritage alone does not predict who lands where.
| Maison | Group | AI Presence | Cultural Desirability | Zone |
|---|---|---|---|---|
| Hermès | Hermès International | 67.2 | 75.6 | Cultural Icons |
| Louis Vuitton | LVMH | 62.7 | 73.9 | Cultural Icons |
| Chanel | Chanel Limited | 55.5 | 81.3 | Cultural Icons |
| Bottega Veneta | Kering | 55.2 | 61.7 | AI-Native Stars |
| Saint Laurent | Kering | 53.0 | 70.6 | Cultural Icons |
| Dior | LVMH | 47.4 | 74.2 | Heritage-Strong / AI-Emerging |
| Gucci | Kering | 46.9 | 63.3 | Awaiting AI Maturity |
| The Row | Independent | 39.7 | 34.2 | Awaiting AI Maturity |
| Miu Miu | Prada Group | 33.6 | 35.8 | Awaiting AI Maturity |
| Alaïa | Richemont | 27.7 | 35.7 | Awaiting AI Maturity |
Bottega Veneta is the Maison the five models agree on most, its cross-model standard deviation of 2.3 the lowest in the panel, its Brand Knowledge score of 87.6 second of ten. They speak about it fluently and in unison. They also, in 33 separate mentions, name a creative director who has moved on, more than any other Maison and 40% of every such error in the study. Dior follows with 27. Together Bottega and Dior account for 60 of the 82 stale-director mentions, and they are precisely the two Maisons whose artistic direction changed most recently and most publicly. The pattern is structural, not anecdotal.
The Maisons AI describes most confidently are often not the ones it describes most accurately.
Hermès leads the panel on AI Presence (67.2) but ranks second on Cultural Desirability, behind Chanel. Chanel leads Cultural Desirability (81.3, the highest in the panel) yet ranks third on presence (55.5). The Maison the models surface most is not the one they hold highest. This is the thesis of the whole index on one map: visibility and desirability are two axes, not one ranking, and the distance between a Maison's place on each is the work that remains.
A Maison can dominate the answer without owning the admiration. Another holds the admiration and surfaces only when asked for by name.
On Heritage Depth the founding Maisons cluster at the top, Chanel 95.1, Saint Laurent 90.5, Dior 89.4, Hermès 87.5, while the youngest contemporaries score zero: the models make no spontaneous heritage attribution to Miu Miu or The Row at all. Alaïa, carrying Azzedine Alaïa's couture legacy, registers a faint 17.9. It is the cleanest separation in the edition. Savoir-faire can be built in a decade and the models will credit it; spontaneous heritage cannot, and they will not invent it. Bottega Veneta is the telling exception, an 80.8 heritage score that reads like a mega-Maison, evidence that AI weighs sustained craft reputation as a heritage signal in its own right.
In fashion, heritage in AI's eyes is still a function of time. The youngest Maisons read at or near zero.
Across the Maison-directed prompts, Google AI Overviews triggered for 52.5% of English queries and 0% of French ones. On the AI Presence axis the AIO layer scores near zero for most Maisons, dragging the channel to the floor for Dior, Miu Miu, Alaïa and The Row alike. The layer closest to mass-market search is structurally absent from luxury fashion, and entirely absent in French. A Maison that reads its AI exposure through English-language Overviews alone will systematically misjudge its presence in its home market.
Absent in luxury fashion, and entirely absent in French.
Most Maisons read consistently across the five models. The disagreement concentrates on a few: Dior (standard deviation 15.0) and Hermès (13.7) are read most differently depending on which model is asked, while Bottega Veneta (2.3) and Gucci (6.9) read almost identically everywhere. For the Maisons where the models split, the narrative is not yet settled, and those are the Maisons for which deliberate source and reference work would move the reading fastest.
Divergence is not noise to be averaged away; it marks the Maisons whose AI representation is still being formed.
Three dimensions compose AI Presence: Product Discovery, Brand Knowledge, Gifting Occasion. Three compose Cultural Desirability: Heritage Depth, Savoir-Faire Authority, Cultural Resonance. Each is scored 0 to 100, model-averaged across all five.
How often, how prominently, how gift-relevantly a Maison surfaces when AI is asked the questions clients ask.
AI is asked to recommend fashion, does the Maison surface?
AI is asked to describe the Maison, does it know enough to answer?
AI is asked for gifts and occasions, does the Maison appear?
How AI describes a Maison once it surfaces, its heritage, its savoir-faire, its place in culture.
AI is asked about lineage, founders, archives, what depth does it carry?
AI is asked about materials, ateliers, techniques, how technically credible?
AI is asked about icons, references, place in culture, how iconic does it sound?
Each score runs from 0 to 100. It is not a percentage and not a sales figure. It reads four signals in every response: how often the Maison appears, how early, in what tone, and with how much detail. Savoir-Faire Authority replaces the Fragrance edition's Olfactory Authority, reading how fluently a model speaks of materials, ateliers and techniques. Gaps under one point sit within the measurement's natural noise: on Brand Knowledge, Dior at 75.6 and Hermès at 75.6 are a tie, not a ranking.
Desirability measures how a Maison is regarded. This audit measures something the desirability score deliberately ignores: whether the model has the facts current. We flag every response that attributes an artistic director the Maison no longer has, anchored to who held the role on the collection date, 2026-06-16. The flag is evidentiary. It never touches a Maison's desirability or valence.
82 mentions · 80 name the wrong person, 2 the right person on a wrong date · Maisons at zero omitted
| Maison | Stale mentions | Wrong person | Wrong date |
|---|---|---|---|
| Bottega Veneta | 33 | 33 | 0 |
| Dior | 27 | 27 | 0 |
| Chanel | 10 | 8 | 2 |
| Gucci | 10 | 10 | 0 |
| Louis Vuitton | 2 | 2 | 0 |
| Σ | 82 | 80 | 2 |
The error is worst where a designer moved between Maisons. The models keep Matthieu Blazy at Bottega Veneta long after he left for Chanel, which is why Bottega carries 33 of the 82 flags. The reading is narrow on purpose: the models are not generally wrong about who runs these Maisons, they lag in one case, when a designer moves from one Maison to another and the old attachment stays. Bottega Veneta (33) and Dior (27) carry the weight. Chanel's ten include the only date errors, where the model names the right designer but mis-dates the arrival. The Maisons with no recent change of director draw zero flags. That is the control: the method reads real lag, it does not manufacture it. Every flag is traceable to its source response, and the raw text is preserved, never corrected.
Google AI Overviews is the one layer that publishes its grounding sources. We classify each as owned (the Maison's own domains), earned (editorial and press) or third-party (platforms and social the Maison does not control). Chat-model sources are model-inferred and excluded from this metric.
Third-party share above 50% is critical. Between 30 and 50% is a watchpoint. Below 30% is healthy.
| Maison | Group | Owned · Earned · Third-party | 3rd-party % | Class |
|---|---|---|---|---|
| Saint Laurent | Kering | 15% | Healthy | |
| Dior | LVMH | 22% | Healthy | |
| Gucci | Kering | 22% | Healthy | |
| Chanel | Chanel Limited | 25% | Healthy | |
| Louis Vuitton | LVMH | 22% | Healthy | |
| Hermès | Hermès International | 26% | Healthy | |
| Bottega Veneta | Kering | 29% | Healthy | |
| Alaïa | Richemont | 31% | Watch | |
| The Row | Independent | 33% | Watch | |
| Miu Miu | Prada Group | 36% | Watch |
Owned shares are reconciled from the edition's underlying export (Google AI Overviews grounding sources); each row sums to 100.
Owned share is in single digits for every Maison, the structural reality of the AI regime: even the most desired Maison sees its authority narrated overwhelmingly by voices it does not direct. The contemporary Maisons carry the highest third-party dependency, and a watch classification with it.
Hermès International
[TODO: confirm savoir-faire figure, dimensions table 76.3 vs profile 76.4]
LVMH
Chanel Limited
Kering
Kering
LVMH
Kering
[TODO: confirm Cultural Resonance, dimensions table 72.3 vs profile 72.4]
Independent
Prada Group
Richemont
A low standard deviation means the five models read a Maison alike; a high one means its AI representation is still contested. Bottega Veneta is read almost identically everywhere (2.3); Dior and Hermès most differently.
| Maison | AI Presence | Cultural Desirability | Cross-model σ | Quadrant |
|---|---|---|---|---|
| Hermès | 67.2 | 75.6 | 13.7 | Cultural Icon |
| Louis Vuitton | 62.7 | 73.9 | 7.2 | Cultural Icon |
| Chanel | 55.5 | 81.3 | 12.8 | Cultural Icon |
| Bottega Veneta | 55.2 | 61.7 | 2.3 | AI-Native Star |
| Saint Laurent | 53.0 | 70.6 | 10.4 | Cultural Icon |
| Dior | 47.4 | 74.2 | 15.0 | Heritage-Strong · AI-Emerging |
| Gucci | 46.9 | 63.3 | 6.9 | Awaiting AI Maturity |
| The Row | 39.7 | 34.2 | 7.4 | Awaiting AI Maturity |
| Miu Miu | 33.6 | 35.8 | 8.3 | Awaiting AI Maturity |
| Alaïa | 27.7 | 35.7 | 13.4 | Awaiting AI Maturity |
Scores model-averaged across Claude, ChatGPT, Perplexity, Gemini and Google AI Overviews. σ is the standard deviation of the Maison's overall score across the five.
The Maisons that lead are those AI both surfaces and reveres. Heritage opens the door; savoir-faire narrative and cultural circulation are what the models reward inside it. Bottega Veneta's mega-Maison heritage score, built on craft reputation rather than age, is the proof.
The 82 stale-director mentions are a currency problem, not a desirability one. They concentrate on the Maisons in transition, and they reward a Maison that grounds its creative-direction narrative at source.
Owned share is in single digits for every Maison. The Maisons most exposed to third-party narration carry a watch classification. Whoever supplies the grounding supplies the story.
Where the five models disagree, the narrative is not settled. Dior and Hermès, read most differently, are the Maisons where deliberate source and reference work would move the AI reading fastest.
AUGMA · The AI Desirability Index · Fashion Edition · June 2026 · Sébastien Pagès · AUGMA / The Augmented Maison.