Artmarket.com builds AI first art platform

Artmarket.com’s majority shareholder structure plans to increase its holding through additional share purchases in the near term, pairing an insider ownership move with the French art-market data company’s decision to rebuild its Artprice business around proprietary artificial intelligence systems.

The company said purchases would occur during authorized trading windows and that required declarations would be filed with France’s Autorité des marchés financiers, or AMF, and published online within the applicable legal deadlines. Artmarket.com did not disclose the intended size of the purchases or a timetable beyond describing them as near-term.

The planned stake increase arrives as Artmarket.com directs spending toward what it calls a vertical AI stack built on the historical auction records, artist indices, images and catalogues held by its Artprice division. The board-approved strategic review has moved the company away from adding isolated AI features to existing products and toward an “AI-first” architecture centered on two in-house systems: Intuitive Art Market® and Blind Spot®.

Artmarket.com said the overhaul requires a minor adjustment to its public rollout schedule, although it maintained that the timing revision does not change its financial trajectory. The company reported continued steady revenue growth at Artprice by Artmarket while citing geopolitical tensions and volatile energy costs as operating challenges.

Artprice rebuild targets an integrated AI platform

The company’s stated approach is to make AI part of Artprice’s underlying production and data infrastructure before releasing new subscriber-facing tools. Artmarket.com said it is restructuring workflows across employees, departments and production units, supported by dedicated AI hardware and edge-computing units.

Its data pipelines for collecting, standardizing and enriching art-market information are also being rewritten around deep-learning standards and proprietary algorithms, according to the company. That work would allow Artprice to deliver a single AI-native environment rather than a collection of separate modules added to a conventional database product.

Artmarket.com identified three technical foundations for the planned platform: standardized big data, deep learning and algorithmic security. In practice, the model depends on whether the company can preserve consistent links between artists, artworks, auction lots, prices, ownership histories and source catalogues across a large and often fragmented market.

That is a more demanding task than producing general-purpose text or image tools. Auction records can contain misspellings, incomplete provenance, title variations, uncertain attribution and inconsistent descriptions between countries. Artmarket.com said its systems must work in more than 40 languages while following an internal rule against translating artwork titles, a constraint intended to preserve the original identifying language of a work.

The company said it will release the subscriber offering after calibrating its internal value chain. No date was provided for the revised public launch.

Blind Spot focuses on price gaps between sales

Blind Spot© is being developed to model price differences between public auction results, Artmarket.com said. The system was initially designed to assess gaps between two public sales by using an artist’s index history and the traceability of auction outcomes.

The company used a Jackson Pollock example to illustrate the type of long-term price comparison it aims to automate, citing a work that sold for $4 million in 1998 and $58 million in 2026. Artmarket.com did not provide further details on the work, sale venues or methodology behind the comparison in the supplied announcement.

A system that identifies price gaps would be useful only if its underlying records distinguish closely comparable works from superficially similar ones. A painting’s size, date, condition, provenance, exhibition history, guarantee arrangements and whether it was offered in a high-profile evening sale can materially affect auction outcomes. Artmarket.com’s strategy places particular weight on data normalization and traceability, rather than relying solely on broad artist-level price trends.

The companion Intuitive Art Market® system is positioned as part of the same AI-first environment, although Artmarket.com provided fewer technical details about its specific functions.

A database-heavy model for AI deployment

Artmarket.com said its proprietary structure includes nearly 180 interconnected databases, along with manuscripts and auction-sale catalogues dating from 1700 to the present. Its databanks contain more than 30 million indices and auction results covering more than 915,300 artists, according to the company.

The firm also said Artprice Images® offers access to 181 million digital images consisting of photographs or engraved reproductions of artworks from 1700 onward. It publishes art-market trends covering 121 countries in 11 languages and operates a marketplace with 9.3 million members, Artmarket.com said.

Data acquisition remains central to the planned systems. Artmarket.com said it works with 7,200 auction-house partners and uses deterministic agents to extract, index and structure information through contractual arrangements and formal partnerships. It is also deploying probabilistic agents for the U.S. market, which the company described as the world’s primary art market.

Deterministic agents follow fixed rules to capture known forms of data, while probabilistic systems infer likely connections or classifications from patterns in large datasets. In an auction context, that could mean helping connect variant artist names, identify repeated appearances of an artwork, or flag records that warrant human review.

Artmarket.com cited estimates from Gartner Group and the Europol Innovation Lab that it characterized as showing “70% uncontrollable synthetic data” as of June 30, 2026. The company used that estimate to argue for the value of historical, certified and structured databases as AI systems become more exposed to machine-generated material online.

Subscription pricing and shareholder base

Artmarket.com said the future subscription offering is expected to cost between €1,600 and $2,500 annually, presented by the company as “€1,600–$2,500.” It also referenced the Financial Times in discussing increasing demand for certified historical databases.

The pricing places the planned service above mass-market consumer AI tools and closer to a specialist professional data product. The commercial test will be whether Artprice can translate its archive into tools that save users time, improve confidence in comparable-sales analysis or uncover useful market relationships without replacing the expert judgment required for high-value art transactions.

Artmarket.com is listed on Eurolist by Euronext Paris. The company said a TPI analysis counted more than 18,000 individual shareholders, excluding foreign shareholders, companies, banks, FCPs and UCITS. It lists Euroclear code 7478, Bloomberg ticker PRC and Reuters ticker ARTF.

Separately, Artmarket.com said Ehrmann has completed a 1,800-page philosophical and scientific essay on AI covering the period from 1987 to the present. The company plans a global digital and print release, with an English edition scheduled before the French version.