Amazon is pursuing an ambitious strategy to strengthen its position in online commerce by building a vast database covering products sold across the global retail market. The initiative, known internally as World Product Graph, reflects a broader effort to address the rise of AI shopping agents while reducing Amazon’s dependence on customers conducting product research through Google, brand websites and third-party review platforms.

Amazon Seeks to Build a Global Product Database

According to an internal document reviewed by Reuters and people familiar with the project, World Product Graph would seek to collect information on virtually every product sold worldwide, including items that Amazon itself does not stock. The initiative would involve brands and retailers continuously contributing product information, while Amazon could also independently crawl other retailers’ websites.

The document says the objective is to make Amazon a more comprehensive destination for product discovery by understanding how products relate to one another. That could allow shoppers to compare models, product versions and colors while receiving recommendations based on their shopping history and other available data.

Amazon has already signed up Coca-Cola and Bayer to contribute product information, while the company also held a CEO-level summit with Procter & Gamble for a trial, according to the document. The companies did not respond to Reuters requests for comment.

AI Shopping Creates a New Competitive Threat

The initiative comes as AI agents increasingly perform research and shopping tasks on behalf of consumers. Amazon executives have become concerned that these systems could direct customers toward competing retailers with more complete or accessible product information, according to people familiar with the matter.

Meta’s Muse shopping agent has reportedly attracted more than 10 million users in the United States and Canada since launching in September. Amazon blocked Muse from its retail website that month and previously sued AI company Perplexity over similar access, highlighting the growing tension between traditional e-commerce platforms and autonomous shopping technology.

Richer Data Could Improve Amazon’s AI Tools

World Product Graph could also address weaknesses in Amazon’s own AI-powered shopping technology. The internal document says Amazon’s Rufus tool, which the company began referring to as Alexa for Shopping in May, has produced answers that are sometimes inaccurate or incomplete.

The document also says customers conduct approximately 83 million searches each week using Rufus and that processing those searches is expensive. A more comprehensive product database could potentially allow Amazon to rely on less expensive AI models while improving the quality and efficiency of responses.

Amazon disputed the reliability of the figures in the document, saying they are significantly outdated and that its data shows Alexa for Shopping is approximately 97% accurate globally. The company also cautioned that the internal document was a draft and that its existence does not necessarily reflect current plans.

Protecting E-Commerce and Advertising Economics

The strategic significance extends beyond product discovery. If Amazon can create the most comprehensive product information system available, AI agents could have greater incentive to rely on Amazon’s database when helping consumers determine what to purchase. That could help preserve customer traffic and support Amazon’s advertising business, which depends heavily on product searches and commercial intent.

The project remains uncertain. Reuters could not determine whether Amazon will implement World Product Graph or whether its reported October 1, 2027 release date is realistic. The initiative could also be modified or abandoned because of financial or other considerations. Nevertheless, the strategy illustrates how AI is changing competition in e-commerce: control over product information may become as strategically important as control over inventory, logistics and customer traffic as shopping increasingly moves toward AI-assisted decision-making.