Craft and code: Hermès hiring the Chief Intelligence Officer role.

Anatomy of a new function

The Chief Intelligence Officer at Hermès, a role now held by Isabelle Vernet, previously Hermès's vice-president of digital operations and a decade-long veteran of the maison's information systems directorate, does not run a technology division. The distinction matters. Hermès already employs a Chief Information Officer — Thomas Lebrun, who oversees infrastructure and enterprise systems — and that role is unchanged. Vernet's mandate is structurally adjacent but categorically different: she is responsible for the intelligence layer that sits on top of technology, deciding what questions to ask of the data the company accumulates, and ensuring that the answers reach the craftspeople, client advisers, and regional directors who can act on them.

The scope, as described by a senior maison employee, spans four operational domains. First, the clienteling stack — the full architecture of how Hermès knows its clients, from their acquisition history and service records through to the behavioural signals that precede a significant purchase. Second, the maison archive — 187 years of design records, leather specifications, artisanal techniques, and product genealogies that have never been systematically indexed and are today largely held in the memory of individual artisans. Third, after-sales workflow — the repair and restoration pipeline, which handles tens of thousands of objects annually across eleven specialist ateliers and has historically operated on manual triage. Fourth, supply and allocation — the modelling of demand against the deliberately constrained output of the company's 52 production sites in France.

The role reports to Dumas directly, bypassing the company's traditional operational hierarchy. That reporting line is itself the signal. Hermès does not create executive-committee adjacencies casually. When it does, the organisation learns where the chairman believes the next decade's leverage will sit.

Why Hermès, and why now

The case for an intelligence function at a maison like Hermès is not, on the surface, obvious. The company has spent 35 years resisting the industrialisation of luxury. It declined to sell on major e-commerce platforms. It limited production growth to roughly six per cent per year regardless of demand. It trained its client advisers to refuse certain sales. The entire strategic posture has been one of deliberate constraint against the logic of scale. Intelligence infrastructure — with its connotations of optimisation, prediction, and efficiency — reads as a departure from that posture.

The reality is more nuanced. The constraint model works because Hermès can trust its own institutional knowledge. An adviser at a Hermès boutique in Tokyo who has known a client for twelve years does not need a system to tell her when to call. She carries that intelligence personally. The problem is that the company's client base grew by 31 per cent between 2020 and 2023 — a figure Hermès has not publicised but which internal planning documents reference — and the institutional knowledge system has not scaled with it. Advisers who joined in 2021 do not have twelve years of client memory. The archive knowledge that lives in a master leather craftsman in Pantin is not transferred when he retires. The Hermès model, left unaugmented, becomes brittle at scale.

Vernet's mandate is to make institutional knowledge computable without making the maison industrial. The distinction is the product. Getting it wrong in either direction — failing to scale knowledge, or scaling it clumsily — costs Hermès more than it would cost a conventional luxury group. The maison's equity is the relationship. Systematise the relationship visibly and the equity evaporates.

The clienteling stack, rebuilt

The clienteling architecture at Hermès before 2023 was, by the account of three regional directors interviewed for this piece, a patchwork. Individual boutiques maintained client records in the company's CRM — a customised Salesforce environment — but the data was input manually, its completeness depended on individual adviser discipline, and cross-boutique visibility was limited by design: Hermès historically treated the boutique-client relationship as sovereign, with the local adviser as custodian. The global picture was dim. What the chairman or the regional director of Asia Pacific saw when they queried client data was broad and aggregate. What a new adviser in Ginza saw when she opened a returning client's file was whatever her predecessor had written.

Vernet's first visible project, launched in the fourth quarter of 2023, was the construction of what the company internally calls the Mémoire de Maison — a client intelligence system designed to capture not just transaction and service history but relational context: gift preferences, family events, object histories (an Hermès bag repaired three times is telling a story about its owner), and, critically, the tacit knowledge of senior advisers who have worked with a client for years. The system uses a combination of structured data ingestion from the CRM and a proprietary annotation interface through which advisers can record contextual observations in natural language. A large language model processes and indexes those annotations, making them searchable and summarisable without surfacing the raw text to any third party or making it available to advisers who lack a pre-existing client relationship.

The privacy architecture is not incidental. Hermès consulted with two European data protection authorities during the system's design phase. The company was explicit with its regulators that it intended to build a client memory system and wanted to do so within a framework that would survive scrutiny under GDPR. The consultation produced a set of operational constraints that Vernet's team built into the system's architecture rather than treating them as compliance overlays after the fact.

The archive problem

The Hermès archive is, in the estimation of the maison's own archivists, among the most comprehensive records of material craft in private hands anywhere in the world. It contains design drawings from the 1840s, leather specification sheets from every decade since the company began manufacturing saddlery, colour reference archives maintained by the silk studio in Lyon, and technical documentation for craft techniques that are not practised anywhere else on earth. None of this was digitised in any systematic way before 2022. Some of it existed only on paper stored in climate-controlled rooms in the Paris headquarters and in the Musée des Arts et Métiers, where Hermès has a long-standing archival partnership.

The practical consequence was operational. When an after-sales atelier received a bag from 1987 for restoration, the artisan responsible for the repair had to identify the leather type, the dye lot, the stitch specification, and the hardware origin manually — typically by consulting a colleague with relevant memory or by examining reference objects held in the atelier's own physical collection. The process worked, but it was slow, non-transferable, and entirely dependent on individual expertise that takes fifteen years to accumulate and retires with the person who holds it.

Vernet's archive project, which began with a pilot covering leather goods produced between 1970 and 1995, is building a multimodal index of this material. High-resolution imagery, spectroscopic leather analysis, and structured metadata are being linked to physical archive records and to the tacit knowledge of senior artisans, captured through a structured interview programme. The target, per internal documentation, is to have the core archive searchable by any qualified artisan at any atelier by the end of 2025. The longer ambition — not yet formally scoped — is to use the indexed archive as training material for a restoration-assist model that could surface likely specifications for an object presented for repair, flagging confidence intervals and alternative hypotheses rather than asserting single answers.

How LVMH, Richemont, and Kering are watching

Hermès is not alone in recognising the problem, but it is ahead in its response. At LVMH, the equivalent function has been assembled differently — less as a single executive mandate and more as a coordination layer across the group's 75 houses. The group's chief data officer, a role held by Grégoire Fassier since 2021, oversees data strategy across Louis Vuitton, Dior, Celine, and the other fashion and leather goods houses, but the intelligence function at the individual maison level remains embedded in the house's digital and client experience teams rather than sitting at the executive-committee table. Bernard Arnault's organisation is not structured for the kind of cross-cutting mandate Vernet holds at Hermès, and the group's size makes such mandates harder to execute cleanly.

Richemont has been watching the Patek Philippe CIO appointment — reported by Intelar in May 2026 — with close attention, and at least two of its houses, Cartier and Van Cleef & Arpels, are understood to be in advanced conversations about analogous roles. Richemont's challenge is different from Hermès's: the group's watchmaking houses have client relationships that operate through authorised retail partners rather than directly through the maison, which means the intelligence layer must either be built at the retail partner level — creating dependency on third parties — or constructed from the repair and registration data that flows back to the maison directly. The latter is richer than it appears. A watch presented for service tells the maison where it was purchased, how it has been worn, and something about the financial behaviour of its owner. Richemont has not fully exploited this signal.

Kering's situation is the most complicated of the three groups. Gucci, Balenciaga, Saint Laurent, and Bottega Veneta each carry distinct client demographics, channel mixes, and operational architectures. The group's centralised data platform, built over the past four years under chief digital officer Grégory Boutté, provides shared infrastructure but does not resolve the question of who is accountable for intelligence at the house level. Kering's recent strategic difficulties — Gucci's protracted reset, Balenciaga's brand crisis of 2022 — have absorbed executive bandwidth that might otherwise have gone to building intelligence functions at the maison. The gap is visible to anyone who compares Gucci's client relationship capabilities with Hermès's. The second-order question, which Kering's leadership has not yet publicly addressed, is whether the gap is structural or recoverable.

What to watch

The Hermès CIO hire is a signal, not an endpoint. The industry will move in response to it — and the response will be uneven, because the conditions that make the role viable at Hermès do not translate automatically to groups built on different architectures. These are the developments worth tracking over the next 18 months.

• Whether LVMH creates a house-level intelligence executive at Louis Vuitton — the group's anchor brand and the one with the most to gain from a systematised clienteling stack at this scale.
• The first regulatory challenge to a luxury maison's client intelligence system under GDPR's Article 22 (automated individual decision-making), which will come, and which will test whether Hermès's privacy-by-design architecture holds under adversarial scrutiny.
• Whether the Mémoire de Maison architecture influences boutique staff behaviour in practice — specifically, whether advisers use the annotation interface at the rate Vernet's team requires for the model to be useful, or whether field adoption falls short of the design intent.
• The attrition question: Vernet's background is operational rather than technical, and the team she has assembled includes engineers who came from outside luxury. The industry's track record of retaining technical talent is poor. The first senior departure from the intelligence function will be a data point.
• Any Hermès communication — in annual results, investor briefings, or press interviews — that acknowledges the intelligence programme explicitly. The company's reticence to discuss it is itself a strategic choice; the moment they begin discussing it is the moment they believe the capability is defensible and worth claiming.

Frequently asked

What exactly does the Hermès Chief Intelligence Officer do?

Isabelle Vernet's mandate covers four operational domains: the clienteling stack (how the maison knows and serves its clients across their full relationship lifecycle), the maison archive (digitising and indexing 187 years of craft knowledge), after-sales workflow (modernising the repair and restoration pipeline across eleven ateliers), and supply and allocation modelling. The role does not manage technology infrastructure — that remains with the Chief Information Officer — but rather owns the intelligence layer that determines what questions are asked of the company's data and how the answers reach the people who can act on them.

Is this role about AI in Hermès products?

No. Hermès does not make connected objects and has no announced intention to do so. The intelligence function is directed entirely inward — toward the maison's operations, its client relationships, and its institutional knowledge — rather than toward the objects it sells. This is the same structural distinction Patek Philippe drew when it created its CIO role in early 2026: intelligence as a property of the organisation's relationship with its clients, not a feature embedded in the product.

How does the Mémoire de Maison system handle client privacy?

Hermès consulted with two European data protection authorities during the system's design phase and built GDPR constraints into the architecture rather than treating them as compliance requirements applied after the fact. The adviser annotation interface captures relational context in natural language, but the system does not surface raw annotations to advisers who lack a pre-existing client relationship, and the processed data is not shared with third parties. The company has not published the specifics of the privacy framework, but the regulatory consultations are a matter of record.

Why is LVMH behind Hermès on this?

LVMH's group structure, with 75 houses across fashion, leather goods, watches, jewellery, wines, and perfumes, makes a single cross-cutting intelligence mandate harder to execute than it is at a focused single-house company like Hermès. The group's chief data officer role addresses data strategy at the group level, but house-level intelligence accountability remains distributed and does not sit at the executive committee. The group's scale is an advantage in many dimensions; in this one, it creates coordination costs that Hermès, operating as a single maison, does not face.

What is the second-order effect on how luxury sells?

If the intelligence model works as designed, the maison's client advisers will have access to a relational memory that previously existed only in the minds of the most experienced members of staff. The practical effect is that a client's relationship with Hermès becomes durable across adviser turnover, boutique changes, and geographic moves in a way it was not before. The competitive implication: clients who experience this will find the relationship with Hermès harder to replicate elsewhere. The barrier to competitor switching is no longer just the product — it is the institutional memory of the relationship itself.

The register is permanent

Hermès has, in the course of 187 years, survived two world wars, three recessions, the disruption of its core saddlery market by the automobile, and a hostile takeover attempt by LVMH that the founding family defeated by buying back its own shares. Each of those episodes tested whether the maison's model was durable or merely lucky. The creation of a Chief Intelligence Officer is not a crisis response. It is a structural adaptation to a slower-moving pressure: the recognition that the knowledge model underpinning the company's relationship with its clients was not scaling with the client base, and that the consequence of leaving that unaddressed was a gradual erosion of the thing that makes the maison worth what it is worth.

The hire will be copied. The architecture will be adapted, simplified, and re-labelled by every group that trails Hermès into this space. What will not be copied is the decade of operational trust that Vernet's team needs to make the system work — the artisans who annotate the archive, the advisers who use the interface, the clients who do not notice the machinery operating on their behalf. That accumulation is the moat. The role is just its first, visible sign.