Amazon develops “World Product Graph,” aiming to map products across the internet for future shopping tools

Amazon is working on an AI project called the “World Product Graph,” a plan described in a recent technology report that centers on creating a structured database of products sold across the internet. The concept, as characterized by the report, is less about tracking individual transactions and more about building a shared, machine-readable “map” of product identities and attributes.

In practical terms, a product graph is designed to connect related products through common characteristics, branding, specifications, and categories, so systems can more easily understand what an item is and how it relates to other items. For large online retailers, this kind of data structure can support faster search, more accurate recommendations, and better answers to product questions, especially when shoppers look across multiple sellers or web stores.

Amazon’s reported approach also indicates a focus on unifying product information from many sources. Instead of treating product pages in isolation, a “world” graph implies cross-site coverage that could normalize the way items are described, reducing the friction created by duplicate listings, inconsistent naming, and variable attribute quality.

The report’s framing positions the initiative as a potential lever for how Americans shop, but it does not provide specific details in the prompt about timing, scale, or whether the effort is already powering consumer-facing features. It also does not disclose how Amazon plans to handle product data quality, duplicates, or disputes over attributes when multiple sellers describe the same item differently.