Patek Philippe's quiet AI-provenance program — Geneva watch market.

The archive and the grading architecture

The substrate for Patek's provenance program is the Maison's archive — the production records the Stern family has maintained since acquiring the company in 1932, the pre-Stern records that go back to the 1839 founding, and the photographic archive of pieces returned to the manufacture for service since the institution of structured service documentation in the late 1960s. The archive is, by any reasonable assessment, the most complete record of any individual luxury watchmaker's production. The Extract from the Archives service, which Patek has operated since 1990, allows owners of references manufactured before 1980 to commission a documentary certificate that the Maison has verified the original specification of the piece against its production records. The service has been the gold-standard provenance document in the secondary market for thirty-five years. It is also, in the period that matters for the new program, a fundamentally manual process: archive staff retrieve original records, cross-reference against the case number and movement number presented by the requester, and issue a paper extract that the requester can present to an auction house or to a dealer.

The AI-provenance program does not replace the Extract service. It supplements it with a layer of computer-vision and forensic assessment that the manual archive process never offered, because the manual process was designed to verify the original specification of a piece rather than to assess the current condition of a piece against its specification. A reference produced in 1955 may emerge from a collector's holdings in 2026 with case dimensions, finishing geometry, dial markings, and hand configurations that have been altered through service intervention, replacement parts, refinishing, or — in the cases that animate the program's commercial logic — deliberate alteration intended to misrepresent the piece's originality. The Extract service confirms what the Maison manufactured. It does not confirm what the piece in the requester's possession currently is. The provenance program is the architecture that closes that gap.

The grading architecture comprises three computer-vision capabilities running on imagery of the candidate piece captured under controlled lighting conditions at the Maison's service department in Plan-les-Ouates, at the Patek Philippe Salon at 41 Rue du Rhône, or at one of the limited number of authorised intake locations the Maison has designated within the program's scope. The first capability is case-wear scoring: an assessment of the geometric correspondence between the case as presented and the case as originally manufactured, calibrated through a reference dataset of pieces whose service histories the Maison can verify completely. The second is dial-originality scoring: a forensic assessment of dial printing, indices, applied markers, lume composition, and the patina patterns that develop on original dials across decades of light exposure. The third is movement-photograph forensics: an assessment of bridge finishing, plate engraving, jewelling configuration, and component originality drawn from movement-side imagery that requires the case-back to be opened by a Patek-authorised technician.

The three capabilities are not equally mature. The case-wear scoring entered production use in the service department in the third quarter of 2024 and has accumulated, by Patek's internal description, the largest reference corpus and the highest scoring confidence. The dial-originality scoring entered production in the first quarter of 2025 and operates with structured human review for any score that falls below a confidence threshold that the Maison has not disclosed publicly. The movement-photograph forensics capability is the most recent — production deployment dates from the fourth quarter of 2025 — and operates only on imagery captured at Plan-les-Ouates or at the Salon, because the lighting and angular requirements for reliable scoring cannot be met in field intake conditions. The capability cascade reflects an internal sequencing decision: case wear was the lowest-risk entry point for the program because case condition is the least disputed dimension of authenticity, dial originality is the most commercially significant because dial replacement is the most common authenticity concern, and movement forensics is the most technically demanding because the reference dataset for movement-side imagery is the smallest. The Maison has built up.

Dealers, auctions, and the integration question

The Geneva grey market for vintage and contemporary Patek references is concentrated within a small number of dealers operating within walking distance of one another on the Rue du Rhône, the Place de Longemalle, and a few side streets in the city's commercial centre. The dealers — names the Geneva watch trade knows without needing them spelled out — operate under a set of conventions that have governed the secondary market for decades: a piece is presented with its papers if available, an Extract from the Archives if the era qualifies for the service, and a verbal warranty from the dealer that the piece is what it is represented to be. The convention has held because the dealers' commercial position depends on the warranty's reliability. A dealer whose representations are later contradicted by a Patek service department's assessment loses the client base that makes the dealership viable. The system has worked.

The introduction of the provenance program has changed the dealers' working position in ways the program's existence has made unavoidable. A dealer who is now selling a 2526 enamel-dial reference to a sophisticated buyer faces a question the buyer did not previously ask: has the piece been assessed against the Maison's provenance program, and if so, what was the result. The dealer can decline to engage the question, but the decline is itself a signal. The dealer can submit the piece to Patek for assessment before selling, but the submission carries a risk: if the program's scoring identifies issues with the dial or case or movement, the dealer's commercial position on the piece is structurally weakened, and the disclosure of the score to the buyer is, in the dealer's commercial reading, effectively forced even if the Maison has not formally required it. The dealers who are best positioned within the new dynamic are those whose stock holdings are dominated by pieces with verifiable service histories — pieces that have passed through Patek's service department in the recent past and that the Maison's records can confirm have remained in their assessed condition. Several of the most established Geneva dealers have moved decisively in this direction during 2025, narrowing their stock to pieces whose provenance can be defended without ambiguity.

The auction-house relationship has evolved more formally. Phillips, in the lead role its Aurel Bacs-era watch department has held since 2014, has been the most active integrator of the program into its authentication workflow. The Phillips watch department's structured process for Geneva and Hong Kong sales now includes, for high-value Patek references, a pre-sale assessment that the Maison conducts on behalf of the auction house at Plan-les-Ouates. The assessment outcomes are not published in the catalogue with the granularity the program's scoring produces — Patek has been clear with the auction houses that the scores themselves are internal documents and not for public disclosure — but the existence of an assessment is signalled in the catalogue copy, and a piece that has been assessed within the program is, by a convention the auction houses and the Maison have developed together, marked with a particular vocabulary in the catalogue description that informed buyers have learned to read. Christie's and Sotheby's have followed Phillips's lead with their own integration protocols, though their adoption pace has been slower and their integration depth, by the description of two specialists at competing houses, less complete.

The asymmetry between Phillips and its competitors carries commercial weight. A consigner who has a complex Patek reference — a 1518 chronograph from a small production year, a 2499 perpetual chronograph in an unusual configuration, a 5970 with a known service history — increasingly directs the consignment toward Phillips because the integration with the program produces stronger market reception. Christie's and Sotheby's have responded by intensifying their own internal authentication protocols and by deepening their bilateral relationships with the Maison's archive function, but the integration depth that Phillips has achieved is, by the description of dealers and auction-house specialists, the benchmark the market currently measures against. The Maison has been deliberate about not appearing to favour Phillips. The Maison has also been deliberate about not extending equal integration depth to every auction house automatically, on the reasoning that the program's value to the secondary market depends on a discipline around how its outputs circulate.

Grey-market pricing and the certificate economy

The financial consequence of the program for the Geneva grey market has been material and measurable. Pieces that have been assessed within the program and that emerge with clean scoring trade at premia to comparable pieces without an assessment, and the premia have widened through 2025. A 5970G in three-register perpetual chronograph configuration with a verified service history and program-assessed provenance traded in October 2025 at the Phillips Geneva auction at a hammer price approximately 14 per cent above the levels at which comparable references without program assessment had traded in the same calendar year. The premium is not uniform across references — it concentrates on pieces whose authenticity has historically been most disputed and on pieces where the Maison's reference corpus is deepest — but the pattern has been consistent enough that the dealers and the auction-house specialists treat the assessed-versus-unassessed distinction as a structural feature of the contemporary market rather than as a temporary signal.

The corresponding consequence at the other end of the distribution has been steeper. Pieces that have been submitted to the program and that emerge with scoring that indicates dial replacement, case refinishing beyond the program's tolerance, or movement-side component substitution have, in the cases where the scoring has filtered into market awareness, traded at meaningful discounts to their pre-program valuations. The discounts have been observable in the secondary trading patterns of specific references that the Maison's archive function has flagged through the program, and the discounts have, in some cases, exceeded 30 per cent for pieces whose representational status before the program had implied originality the assessment did not confirm. The market is, in effect, re-pricing the certificate economy in real time as the program's outputs accumulate.

The certificate-trust dynamic that has historically organised the secondary market has, through this re-pricing, undergone a structural shift the program's architects appear to have anticipated and the broader market is still adjusting to. The Extract from the Archives, which has been the dominant provenance instrument for thirty-five years, was a document that the Maison issued based on its production records. Its limitation was that it spoke to original specification and not to current condition. The program's assessment outputs speak to current condition relative to original specification, and the assessment is, in the trust hierarchy that the market is constructing around the program, the higher-status instrument. The Extract remains relevant. The assessment is becoming the document that high-value buyers expect to see, and the absence of an assessment on a high-value piece is read as a signal even when the Maison's archive function has not made an explicit statement about the piece.

The asymmetric information problem that has bedevilled the secondary market for decades — the problem that the seller knows more about a piece's history than the buyer can verify — has been partially resolved by the program's architecture. The resolution is not complete: the program's scoring is held internally by the Maison and circulates to the market through structured channels rather than as a published score attached to a piece's commercial identity. The asymmetry has shifted, however, from a buyer-seller asymmetry to a more complex three-party arrangement in which the Maison sits as a referee whose assessment can be invoked but whose outputs are not unilaterally disclosed. The dealers and auction houses have absorbed the new architecture by adapting their commercial conventions. The buyers have absorbed it by developing the vocabulary to ask informed questions about whether and where a piece sits within the program's process. The market is, in this sense, more efficient than it was, but the efficiency comes from infrastructure the Maison controls rather than from neutral certification — a distinction that matters more to some buyers than to others and that has become part of the contemporary politics of the Patek market.

The disclosure posture: why Patek will not publish

The Maison's reluctance to discuss the program publicly is not an oversight or a temporary positioning. It is a deliberate posture that Thierry Stern, Patek Philippe's president since 2009 and the fourth generation of the Stern family to lead the house, has been explicit about in private conversations with senior figures in the trade and in his published statements about the Maison's relationship to communication and discretion. Stern's view, communicated in a series of interviews he gave to the trade press in 2024 and 2025, is that the Maison's commercial proposition depends on a relationship to its clients and its objects that is mediated through institutional discretion rather than through visible disclosure of operational practice. The program is an internal tool that the Maison operates in service of the market's health. It is not a marketing instrument and it is not, in Stern's reading, an instrument whose existence the Maison needs to celebrate.

The disclosure posture has practical consequences for how the program operates. The Maison does not publish the scoring thresholds the program uses, does not publish aggregate statistics about the program's outputs, does not disclose which dealers and auction houses have access to the structured integration protocols, and does not provide buyers with mechanisms to verify directly whether a piece has been assessed outside the channels the Maison itself controls. The opacity is intentional. It creates a market dynamic in which information about the program circulates through trusted relationships — dealers known to the Maison, auction houses with established integration, collectors whose long-standing client relationships qualify them for access — rather than through public channels that the Maison cannot govern. The market that emerges is structured around relationship-mediated trust, and the relationship-mediated trust is the architecture the Maison considers most compatible with its commercial proposition.

The contrast with the disclosure posture of contemporary auction-house authentication services — particularly the protocols that Christie's and Sotheby's have published for their internal authentication work — is instructive. The auction houses operate in a regulatory environment that demands published methodology, transparent assessment criteria, and disclosure of authentication outcomes to consigners and buyers within structured timeframes. Their internal authentication processes are necessarily more public than Patek's program, because the auction houses are intermediaries whose commercial credibility depends on the visible robustness of their authentication. The Maison is not an intermediary. It is the manufacturer whose archive is the underlying truth that any authentication ultimately references. The asymmetry of position justifies, in Stern's posture, an asymmetry of disclosure. The Maison can afford opacity in a way an auction house cannot, and the opacity itself contributes to the Maison's mystique and to the value of access within the relationship-mediated economy the Maison cultivates.

The disclosure posture has critics within the trade. A handful of independent collectors and a few specialist publications have argued that the program's opacity creates a market structure in which the dealers and auction houses with privileged access to the Maison's outputs hold an information advantage over the broader buyer base, and that the advantage is incompatible with the program's nominal purpose of improving market health. The Maison's response to this critique, communicated through indirect channels because Patek does not engage directly with trade journalism on operational matters, has been that the program's purpose is to support the market through institutional channels rather than to flatten the market into a fully transparent information environment, and that the Maison's reading of luxury commerce is that institutional channels are how trust is built and disclosed transparency is how trust is commoditised. The critique is, in this reading, applying mass-market expectations to a market that operates by other conventions. The Maison's position has not moved.

How the program fits into Patek's broader posture

The provenance program does not exist in isolation. It sits within a broader pattern of operational investment by Patek across the past five years that has, by the Maison's standards, been substantial and deliberate. The Maison has expanded its service capacity at Plan-les-Ouates, added authorised intake locations in markets where its installed base concentrates, deepened its archive digitisation programme so that the Extract from the Archives service can extend further into the post-1980 production that was previously outside the service's scope, and built up a research function within the company that the Stern family treats as a long-term investment in the Maison's institutional knowledge rather than as a productivity instrument. The provenance program is one expression of this pattern. It is not the only one.

The pattern reflects a strategic posture that Thierry Stern has articulated publicly across multiple interviews: the Maison's commercial proposition is built on the durability of its objects and the durability of its relationships, and the durability of both requires institutional investment in the infrastructure that maintains them across generations. The market for vintage Patek references is, by Stern's reading, not a secondary market in the sense that the term is used in the broader luxury trade. It is the active commercial environment in which the Maison's objects continue to live across decades and centuries, and the Maison's investment in the health of that environment is investment in the value proposition the Maison sells when it produces a new piece. A client who buys a 5236P perpetual calendar today and pays the Maison's current price is paying, in part, for the institutional assurance that the same reference will be supportable, serviceable, and authenticable in 2070 or 2100. The provenance program is part of that assurance.

The contrast with Audemars Piguet's commercial posture, which Intelar has reported on extensively in coverage of the AP Chief Intelligence Officer programme, sharpens the reading of Patek's choices. AP has built a direct retail estate and a client residency model that organises its commercial infrastructure around the relationship between the maison and the contemporary buyer. Patek has invested instead in the institutional infrastructure that supports the object across its long life — the archive, the service department, and now the provenance program — on the reading that the relationship between the maison and the client is mediated, ultimately, through the object's durability. Both postures are defensible. They are not the same, and the difference produces different operational priorities. AP's chief intelligence officer programme builds a client coherence layer across a global boutique network. Patek's provenance programme builds an object coherence layer across a global secondary market. The maison's resource allocation reflects which coherence it considers the durable foundation of its proposition.

The provenance program is also, in Patek's strategic reading, a defence against a specific risk the broader market has not yet absorbed: the emergence of generative AI tools that can produce convincing photographic forgeries of vintage pieces at quality levels approaching expert detection. The risk is not the production of physical forgeries — that has been a feature of the watch market for decades and the Maison's manufacture-level inspection has detected physical fakes consistently. The risk is the production of photographic documentation that purports to show provenance, service history, or authenticity at a level of visual convincingness that the previous generation of forgery techniques could not achieve. The program's computer-vision capabilities, by the description of two people familiar with the Maison's internal rationale, were structured in part to operate against this risk: to assess physical pieces against authenticated visual references, rather than to assess submitted photographs against expectations, because the photograph itself can no longer be treated as a neutral source. The architectural choice was made in 2023 and has, in retrospect, proved prescient. The Maison's reading of where the risk to authenticity was heading was earlier than most of the trade's. The program is what an early reading looks like in operational form.

What to watch

Patek's provenance program is approximately twenty months into production deployment and has become, by the description of dealers and auction-house specialists in the trade, a structural feature of the secondary market for the Maison's references. The next twelve months will determine how the program's architecture extends, how its disclosure posture adapts to market pressure, and how competing maisons respond.

• Whether the program's scoring outputs become incorporated formally into auction-house catalogue copy beyond the implicit signalling currently in place; the convention of marking program-assessed pieces with vocabulary that informed buyers can read has held during 2025, but the pressure from buyers and consigners for more explicit disclosure is increasing, and the Maison's posture on whether to permit publication of summary scoring will be the most consequential disclosure decision the program faces in the coming year.
• Whether competitor maisons in the haute-horlogerie segment construct comparable provenance programs; Audemars Piguet has not announced one, Vacheron Constantin is understood to be in early scoping under its Chief Intelligence Officer programme, and Patek's program is, for the present, the only architecture at scale, but the competitive pressure on the other independents and on the Richemont-owned grand maisons will intensify if the program's market influence continues to widen through 2026.
• Whether the program's reference corpus is extended to incorporate post-1985 production at the scale the pre-1985 corpus currently covers; the Extract from the Archives service has historically been most robust for the early- and mid-twentieth-century production, and the program's scoring confidence is correspondingly highest for older pieces, but the contemporary secondary market increasingly trades references from the 1990s and 2000s whose provenance questions are different in character, and the program's reach into this newer corpus will determine its commercial relevance to the bulk of secondary trading volume.
• Whether the disclosure posture withstands an enforcement action by a national consumer-protection regulator that takes the view that the program's opacity creates an information asymmetry incompatible with consumer protection in high-value secondary transactions; the European regulatory environment for luxury secondary markets has tightened through 2024 and 2025, and the Maison's disclosure architecture has not been tested by a formal regulator inquiry yet.
• Whether the integration depth that Phillips currently holds extends symmetrically to Christie's and Sotheby's, or whether Patek's strategic preference is to maintain a tiered integration architecture that rewards the auction house whose authentication culture most aligns with the Maison's; the answer will determine whether the program becomes a market-wide standard or a Phillips-favouring instrument, and it will shape the broader competitive dynamic among the three top-tier watch auction houses for the rest of the decade.

Frequently asked

What is the difference between the Extract from the Archives and the new provenance program?

The Extract from the Archives, operating since 1990, is a documentary certificate that Patek issues based on its production records, confirming the original specification of a piece manufactured before 1980 against its case number and movement number. The provenance program supplements the Extract with a computer-vision and forensic assessment of the current physical piece against the original specification: case-wear scoring, dial-originality scoring, and movement-photograph forensics. The Extract speaks to what was manufactured. The program's assessment speaks to what the current piece actually is, and addresses the gap that the manual archive process never closed.

Where can a piece be submitted for assessment, and is the program open to all owners?

Submission is currently limited to the Maison's service department at Plan-les-Ouates, the Patek Philippe Salon at 41 Rue du Rhône in Geneva, and a limited number of authorised intake locations the Maison has designated. The movement-photograph forensics capability operates only on imagery captured at Plan-les-Ouates or the Salon, because the angular and lighting requirements cannot be met at field intake locations. The program is open to owners of pieces that fall within its production-era scope, but in practice access has been mediated by the Maison's existing relationships with authorised dealers and with auction houses, and direct submission by individual owners without an established channel has been rarer than channel-mediated submission.

How does the program integrate with Phillips, Christie's, and Sotheby's, and why is Phillips ahead?

Phillips's watch department, under Aurel Bacs's long-standing leadership, has been the most active integrator: for high-value Patek references entering Phillips's Geneva and Hong Kong sales, the Maison conducts a pre-sale assessment at Plan-les-Ouates on behalf of the auction house. The assessment outcomes are not published in catalogue copy with the program's scoring detail — Patek treats the scores as internal documents — but the existence of an assessment is signalled through a particular vocabulary in the catalogue description. Christie's and Sotheby's have followed with their own integration protocols, but at slower adoption pace and less complete integration depth. Phillips's lead reflects the closer cultural alignment between its authentication culture and the Maison's posture, not formal exclusivity.

What is the measurable impact on grey-market pricing for assessed versus unassessed pieces?

The pricing impact concentrates on references whose authenticity has historically been most disputed and on references where the Maison's reference corpus is deepest. A representative data point: a 5970G three-register perpetual chronograph with verified service history and program-assessed provenance traded at the Phillips Geneva October 2025 sale at a hammer price approximately 14 per cent above comparable unassessed references in the same calendar year. The corresponding effect at the other end of the distribution has been steeper: pieces whose program scoring indicates dial replacement or other authenticity issues have, where the scoring has reached market awareness, traded at discounts of 30 per cent or more relative to their pre-program valuations.

Why does the Maison refuse to publish the program's methodology or aggregate outputs?

The disclosure posture reflects Thierry Stern's strategic reading of the Maison's commercial proposition: that Patek's commercial relationship with its clients and its objects is mediated through institutional discretion rather than visible operational disclosure, and that public publication of methodology and outputs would convert the program from an instrument of relationship-mediated market support into a commoditised authentication service of the kind operated by neutral third-party authenticators. The Maison's reading is that the value of the program to its clients and to the durability of its objects depends on the program operating within relationships the Maison knows and governs, not as a public information service. The posture has critics within the trade; the Maison's position has not moved.

How does the provenance program relate to the threat of generative AI in producing photographic forgeries?

The program's computer-vision capabilities were structured, in part, to operate against the risk that generative AI tools would produce convincing photographic documentation of provenance, service history, or authenticity at quality levels approaching expert detection. The architectural choice — to assess physical pieces against authenticated visual references in controlled-imaging environments, rather than to assess submitted photographs against expectations — was made when the Maison's internal research function identified the photographic-forgery risk as emerging in 2023. The program is, in this reading, both a response to the existing authenticity-fraud problem in the secondary market and a structural defence against a forgery vector the broader market has not yet absorbed.