Amazon Drafts Plan to Build Global Product Database to Counter AI Rivals

Amazon is developing an internal initiative known as the World Product Graph, a comprehensive strategy aimed at aggregating data on virtually every product sold globally, regardless of whether the company currently stocks the item. According to an internal document reviewed by Reuters, the plan seeks to transform Amazon into a definitive destination for shopping by providing deep insights into product relationships and specifications. The move is designed to challenge Google's dominance in online search and counter the expanding e-commerce operations of rival Walmart.

The document, which outlines the rationale and business case for the new service, addresses the rising prevalence of autonomous AI shopping agents. These software tools can complete purchasing tasks on behalf of users, and the internal memo notes that Amazon's own AI-driven search agent, Rufus, frequently provides inaccurate information. By centralizing data, Amazon hopes to reduce reliance on external sources and improve the accuracy of its recommendations.

Reuters reported that the document includes an aspirational future release date of October 1, 2027, though it is unclear if this timeline is realistic. Amazon stated in a press release that while the document reflects brainstorming ideas, the company is constantly working to make its product catalog more comprehensive and accurate. The company also noted that some usage figures in the draft are outdated and that its current data suggests the global accuracy of its shopping assistant is roughly 97 percent.

The implementation of the World Product Graph would involve crawling retailer websites to gather product information. While some brands have already agreed to contribute data continuously, including Coca-Cola and Bayer, Amazon may also independently scrape data from other sites. The company held a high-level summit with Procter & Gamble to trial this approach, though neither company responded to requests for comment regarding the specific terms of data sharing.

Internal staff are currently being reorganized to support the project, which falls under the Selection & Catalog Systems unit led by Vice President Robert Tekiela. The initiative is part of a broader effort to address the limitations of existing search capabilities. By having more complete data, Amazon aims to help customers identify the most recent models or specific colorways of electronics and guide them toward products that best fit their personal preferences and shopping history.

The project also aims to solve efficiency issues with Amazon's AI search engine, Rufus, which was rebranded as Alexa for Shopping in May. The internal document describes the current operation of Rufus as expensive, with users conducting approximately 83 million searches weekly. With a more robust database, Amazon could utilize less costly AI models to process these queries more efficiently.

Strategically, the World Product Graph is positioned to defend against third-party AI agents that might direct traffic to competitors. CEO Andy Jassy has previously downplayed the immediate impact of AI agents on Amazon's advertising business, stating that few customers use them for significant shopping volumes. However, executives remain wary of rivals like Meta, whose Muse agent has gained over 10 million users in the US and Canada since its September launch.

Amazon has taken steps to protect its platform, including blocking Meta's Muse agent and suing AI firm Perplexity over similar access. By owning the most extensive storehouse of product data, Amazon intends to ensure that AI agents working for shoppers are compelled to source products directly from its site. This would help protect its growing advertising revenue and solidify its position as a one-stop shop for all consumer needs.

Employees have described the project as an expansion of existing services like Buy for Me and Shop Direct, which allow customers to purchase items from other retailers through Amazon. The goal is to provide the missing context in online shopping, helping users understand how different products relate to one another and which specific item is the right choice for their individual needs.