On March 30, 2026, the inaugural 2026 Procurement & Supply Chain Salon, hosted by the Public Procurement Branch of the China Federation of Logistics & Purchasing and co-organized by JD Industrials, was held in Beijing. The salon focused on the theme of an "E-commerce Procurement Price Index for State-Owned Enterprises." Nearly one hundred representatives from relevant government ministries, numerous central and state-owned enterprises, e-commerce platforms, and supply chain service institutions attended the meeting to explore new pathways for the digital and intelligent transformation of procurement and price compliance control in state-owned enterprises. Participants engaged in heated discussions on key issues such as ensuring the fairness of price indices, unifying product standardization, and the deep application of AI technology. They proposed establishing an industry price reference system supported by multi-dimensional data that considers quality, service, and regional differences, thereby promoting a shift in procurement focus from "price orientation" to "value orientation." Attendees unanimously agreed that establishing a widely recognized industrial product procurement price index is crucial for enhancing the compliance, transparency, and efficiency of procurement. It was suggested that industry associations take the lead in promoting multi-party data co-construction and sharing.
Previously, JD Industrials announced the external release of its industrial product price index. This index is built upon the company's proprietary AI model, JoyIndustrial, and has been rigorously tested internally, demonstrating excellent price fairness. This initiative directly addresses the demand for authentic, credible, and publicly verifiable industrial product procurement prices, laying a solid foundation for the high-quality development of the industrial supply chain industry and for compliant procurement by enterprises, particularly state-owned enterprises. Representatives noted that JD Industrials' extensive product coverage, strong price credibility, and high reliability have made it a core benchmark for price verification in the e-procurement systems of many companies. JD Industrials' willingness to open its valuable data reflects a mindset of open collaboration and mutual benefit, which is expected to drive high-quality development across the entire industry.
**Industrial AI Model Accelerates Mercator Standard Product Library Development** During the thematic sharing session, representatives from JD Industrials, State Grid E-Commerce, and China Energy Conservation shared practical experiences. A representative from JD Industrials stated that constructing an industrial product price index faces significant challenges due to the vast variety of product categories, diverse specifications and parameters, and inconsistent naming conventions, which hinder price comparability and the scientific robustness of the index system. By leveraging its self-developed JoyIndustrial AI model, JD Industrials is advancing industrial product standardization through the Mercator Standard Product Library. Innovations in large model algorithms enable accurate identification of equivalent products, providing a solid foundation for price governance.
JD Industrials' Mercator Standard Product Library utilizes technologies like deep learning to analyze information related to standardized products, including names, models, specifications, parameters, images, and manufacturer details. By integrating product information from the supply chain and the knowledge of industry experts, and through extensive data cleansing and extraction, a unified system of product parameters with consistent definitions and specifications has been established. This effectively reduces communication and transaction costs during the procurement process. The application of the JoyIndustrial AI model has significantly accelerated the development of the Mercator library. It was revealed that, with the model's assistance, the volume of product data organized by JD Industrials in the past year has exceeded the total from the previous five years combined.
**Precise "AI Standard Product Translation" Solves Equivalent Product Identification Challenges** Procurement of industrial products often involves significant challenges in sourcing and price comparison due to imprecise product descriptions. Different users frequently use colloquial or habitual terms for the same item, such as "screwdriver" or "turn-screw," making it difficult to accurately identify the target product. Building upon the standard product library, large model technology has brought substantial improvements in both efficiency and effectiveness for equivalent product identification, enabling a shift from fuzzy to precise matching.
Even when user product descriptions are vague, or when brand and model names are written non-standardly or incorrectly, the JoyIndustrial AI model can identify the correct item by drawing on its dataset of tens of millions of industrial products and extensive industry practice data, performing precise "AI standard product translation." Based on its algorithms, JD Industrials can quickly identify the same product listed under different names across multiple platforms, accurately matching identical items, resolving issues of multiple codes for one product, and laying the groundwork for subsequent like-for-like price comparisons. For example, from a user's description like "10-piece set of open-end wrenches" and accompanying images, the system can translate this into a precise standard description such as "**[Brand] 10-Piece Full Polished Double Open End Wrench Set *1 / 6 inch / 5.5-32mm**," overcoming problems caused by vague descriptions and variations to enable truly accurate and effective price comparisons.
**JoyIndustrial Model Integrates the Value of Data and Scenarios** Behind the development of the industrial product library and equivalent product identification capabilities, the JoyIndustrial AI model plays a critical role. A JD Industrials representative emphasized during the分享 that "continuously increasing investment in the research and development of industrial AI models is a fundamental requirement for solving the challenge of equivalent product identification and building a reliable price index." He also highlighted the value of industrial models in practical applications: due to the specialized and unique nature of industrial data, general-purpose large models struggle to achieve accurate product identification. Furthermore, companies often prefer to train and apply models within their private domains to ensure information security, and they benefit from using relatively smaller, more specialized models that offer higher execution efficiency and lower operational costs.
In 2025, JD Industrials launched JoyIndustrial, the industry's first dedicated industrial AI model. Trained on data from over 97.7 million industrial product SKUs accumulated through JD's long-term service to the industrial sector, combined with practical experience from more than 40 sub-sectors, the model leverages deep industry insight and professional expertise. It enables precise intelligent decision-making and process optimization, leading to substantial reductions in supply chain costs and multiplicative improvements in operational efficiency. The intelligent capabilities output by JoyIndustrial cover multiple stages including product management, transactions, and fulfillment. Coupled with a series of application-specific agents launched for key business scenarios, it achieves an organic integration of massive vertical "data" and specific industry "scenarios," has been robustly implemented both internally and externally, and has created clear, demonstrable value.
Peng Xinliang, Secretary-General of the Public Procurement Branch of the China Federation of Logistics & Purchasing and Director of the Procurement Committee, stated that compliance and price transparency in e-commerce procurement have become focal points for the industry. Leveraging the standard product libraries and data resources of leading companies like JD Industrials, and collaborating with various stakeholders to build a neutral, transparent third-party price reference mechanism, will advance the construction of a price sharing platform for state-owned enterprise e-procurement and facilitate the compliant and intelligent upgrade of state-owned enterprise procurement.
Comments