Luxury Retail AI in 2026: LVMH's AI Factory, Kering's Labs, Farfetch's Collapse, and Clienteling at Scale

Luxury retail is the AI vertical with the most distinctive constraints. The customer relationship is genuinely high-touch, the brand-perception cost of an AI misstep is asymmetric, and the data is unusually concentrated in a small number of high-value relationships rather than spread across millions of transactions. The major luxury groups — LVMH, Kering, Richemont, Hermès, Prada — have spent the 2022-2026 window building AI investments that respect those constraints, with mixed publicity and uneven success. Alongside the groups, the pure-play luxury e-commerce platforms — Farfetch, Mytheresa, Net-A-Porter, MatchesFashion — produced one of the largest single category collapses in recent retail history.

This is the practical sort of what’s shipping in luxury retail AI, what isn’t, and what the Farfetch outcome means for the category.

The Farfetch collapse and Coupang acquisition#

Farfetch — the London-listed luxury marketplace founded by José Neves — went from a 23 billion USD peak market capitalisation in 2021 to a financial collapse in late 2023 that ended in Coupang’s emergency acquisition for what amounted to a fraction of peak value. The collapse exposed several structural problems the marketplace model had quietly carried: thin marketplace economics on luxury goods, expensive customer acquisition relative to lifetime value, the unfortunate timing of the YOOX Net-A-Porter (YNAP) acquisition from Richemont that proved unfundable, and the broader luxury slowdown in 2022-2023.

The Coupang acquisition, closed in early 2024, brought Farfetch under the Coupang umbrella with significant operational restructuring. The 2024-2025 rebuild focused on the brand and white-label e-commerce services for luxury houses rather than the consumer marketplace expansion narrative of the Farfetch peak era. Sequoia, Tencent, and the other Farfetch backers absorbed the loss.

The lesson the category took: luxury e-commerce at marketplace scale requires either a meaningful direct relationship with the houses (a constraint that limited Farfetch’s catalogue access) or the operational discipline of a Mytheresa or Net-A-Porter that the marketplace model never imposed. The AI investments in personalisation and recommendation, while real, could not compensate for the underlying commerce model.

LVMH’s AI Factory#

LVMH — the largest luxury group — operates an internal AI Factory that supports the maisons (Louis Vuitton, Dior, Tiffany, Bulgari, Sephora, Moët Hennessy, and the rest of the portfolio) with shared AI infrastructure and embedded teams. The 2024-2025 disclosures described over 1,000 AI use cases across the group, ranging from supply-chain forecasting to customer-service automation to creative-team productivity tooling.

The LVMH model is illustrative for any conglomerate-scale AI strategy. The shared platform handles the infrastructure (data lake, ML platform, foundation-model access), the embedded teams handle the maison-specific use cases (Sephora’s recommendation needs differ materially from Bulgari’s clienteling needs), and the governance layer handles brand-safety review before any customer-facing AI ships under a maison brand.

The publicly discussed wins include Sephora’s AI-driven recommendation and product-matching tools, Louis Vuitton’s supply-chain forecasting improvements, and the broader productivity gains across back-office functions. The under-publicised work includes the careful, multi-quarter review of any customer-facing AI before it ships, and the meaningful share of pilots that never made it to production because the brand-perception risk did not justify the operational gain.

Kering’s AI labs#

Kering — Gucci, Saint Laurent, Bottega Veneta, Balenciaga, and the rest of the group — operates a more distributed AI structure than LVMH, with maison-level AI teams and a central infrastructure layer. The 2024-2025 communication emphasised the AI investment in supply-chain visibility (a meaningful operational gap for Gucci specifically during the 2022-2023 slowdown), the rebuild of Gucci’s customer relationship management, and the AI-augmented productivity work for creative teams.

Kering’s strategic position in 2026 is complicated — Gucci’s growth slowdown, the sale of the beauty division to L’Oréal, the broader luxury market softness — and the AI work sits inside that complicated context rather than as the headline story.

Mytheresa and the disciplined e-commerce model#

Mytheresa (Munich-based, Neiman Marcus-owned through 2024 before the broader restructuring, currently independent and acquiring YNAP from Richemont through the late-2024 announced deal) became the demonstrated counterexample to the Farfetch collapse. The Mytheresa model — curated rather than marketplace, direct house relationships rather than third-party seller aggregation, and operational discipline on customer acquisition cost — produced sustained profitability where Farfetch could not.

The Mytheresa AI investments are deliberately less publicised but substantively focused on the personalisation of the curated assortment, the high-touch client communications (the “personal shopper” experience scaled with AI assistance), and the operational efficiency of inventory and returns management. The disciplined approach to the customer base — Mytheresa’s top customer cohort produces a disproportionate share of revenue and the AI prioritises that cohort’s experience over volume metrics — is the operational pattern that distinguishes the survivor from the casualty.

Net-A-Porter and the YNAP transition#

The YNAP business — Net-A-Porter, Mr Porter, The Outnet — operated under Richemont through 2024 in an extended sale process, with the Mytheresa acquisition announced late 2024 to consolidate the major curated luxury e-commerce platforms under one operator. The 2024-2026 transition has been complex; the AI investments under Richemont’s ownership focused on personalisation and clienteling, and the post-acquisition integration with Mytheresa’s stack is the 2026-2027 operational work.

In-store clienteling AI#

Clienteling — the relationship management between sales associates and high-value clients — is the AI category where luxury retail has invested most distinctively. The associate’s tablet (Sephora, Louis Vuitton, Tiffany, Bulgari, Gucci, Burberry, Coach, Dior, Hermès — varied by maison) increasingly carries an AI-augmented client profile: purchase history, stated preferences, life events, the associate’s prior notes from past visits, recommended items in stock that match the client’s pattern, and increasingly the AI-suggested talking points for the visit.

The clienteling vendors include Salesforce Service Cloud at the larger maisons, NewStore for boutique-format operators, Tulip for the in-store experience, and a tail of regional integrators. The AI investments are typically embedded in those platforms rather than purchased as separate AI products.

The clienteling success story in luxury is the same as the personalisation story elsewhere — the value comes from the data layer, the integration with inventory and CRM, and the associate base’s commitment to using the tool.

Counterfeit detection — Entrupy and Authentic Vision#

The counterfeit-detection vendors became operationally significant for luxury retail and resale alike. Entrupy (US-based, founded 2012) uses microscope-level imagery and AI classification to authenticate handbags and accessories across the major luxury houses; the technology is used by The RealReal, Vestiaire Collective (covered in the resale post), Rebag, and increasingly by the maisons themselves for grey-market and gift-card fraud detection.

Authentic Vision (Austrian-origin) takes a different angle — the brand embeds a unique cryptographic signature in the product at manufacture, and the consumer or merchant verifies authenticity via mobile scan against the signature. The integration is upstream at the brand rather than downstream at the resale platform, and the partnership list includes several luxury houses on selective categories.

The counterfeit-detection AI is one of the few areas in luxury retail where the use case is genuinely novel rather than a refinement of existing patterns — and where the regulatory backdrop (the EU’s anti-counterfeit directives, the platform liability shifts under the Digital Services Act) makes the technology operationally necessary rather than discretionary.

The brand-perception constraint#

The asymmetric brand-perception risk in luxury AI is the single largest operational constraint. A customer-service chatbot that produces a mildly wrong answer at a mass retailer creates marginal annoyance; the same chatbot at a Hermès or Bulgari touchpoint creates a brand-perception incident the maison will reasonably treat as unacceptable. This drives several patterns:

The customer-facing AI is more conservative than the equivalent at mass retail — extensive human review, narrower scope, faster handoff to a human associate when the AI signals uncertainty.

The associate-facing AI is more aggressive — the AI augments the associate’s capability without speaking directly to the client, with the brand-perception risk filtered through the associate’s judgement.

The back-office AI — supply chain, finance, creative-team productivity, manufacturing operations — operates with less constraint and absorbs the majority of the practical AI investment.

This pattern is structurally different from the mass-retail AI playbook and produces a different vendor-evaluation framework. Luxury groups buying AI evaluate brand-safety, governance, and human-handoff design at least as carefully as model quality.

What we recommend in 2026#

For a luxury operator scoping an AI program:

• Back-office and supply-chain AI before any customer-facing investment — the LVMH and Kering patterns deploy here first because the brand-perception risk is low.
• Associate-facing clienteling AI as the second-stage investment — the tablet-in-the-boutique pattern is well-proven, and the value flows through the associate’s existing judgement.
• Customer-facing AI as a third-stage investment with multi-quarter brand-review process — start with low-risk surfaces (order status, basic product information) before any conversational commerce ambition.
• Counterfeit detection as an operational requirement — Entrupy, Authentic Vision, or the equivalents now belong in the maison’s standard infrastructure.
• Curated rather than marketplace as the e-commerce model — the Mytheresa pattern over the Farfetch pattern.

Related reading#

• Second-hand marketplaces and authentication
• E-commerce personalization platforms
• Returns optimization

Where pdpspectra fits#

We help luxury operators stand up the data layer behind clienteling, supply-chain visibility, and the back-office AI — POS, CRM, inventory, and customer-history pipelines plus the integration glue between the maison’s systems and the group-level AI platform. Our AI and LLM integration practice handles the model side; the data engineering team handles the operational layer.

---

Luxury AI is about brand-safe deployment, not just model quality. If you are scoping a clienteling rebuild, a supply-chain visibility program, or counterfeit-detection integration, tell us about the maison.