Amazon's 'World Product Graph' Would Index Every Product Online

Amazon has an internal plan to catalog every product sold anywhere on the internet — including products it doesn't carry — and surface that data on its own site, according to a Reuters report. The project carries an October 1st, 2027 release date in the internal document Reuters reviewed.

The plan has a name: the "World Product Graph." The goal is straightforward. Amazon wants to stop shoppers from leaving its site to research products on Reddit or Google, and it wants people who search for items sold elsewhere to still land on Amazon's pages. The bet extends to AI agents too. If Amazon holds the largest stockpile of structured shopping data, AI shopping bots may default to its index over anyone else's — so even bot-assisted buyers end up on Amazon.

What does the project change for shoppers?

In practice, the World Product Graph would make Amazon a product research destination rather than purely a store. Instead of scrolling Reddit threads for whether a vacuum is worth the money, a shopper could find consolidated product details — even for items Amazon doesn't sell — inside Amazon's interface.

The groundwork already exists. Links to outside retailers' products have started appearing on Amazon through the "shop direct" feature the company launched last year. That feature lets Amazon monetize demand it can't fulfill itself, and the World Product Graph would deepen the pattern.

Why does Amazon need more data for its AI?

The project also targets a problem inside Amazon's own AI shopping assistant, Alexa for Shopping — previously called Rufus. The internal document is blunt about its shortcomings, stating: "Many Rufus answers are factually wrong or incomplete."

There's a cost problem too. Running users' Amazon searches through the AI is expensive, per the document. More data in the system is supposed to fix both issues at once: better accuracy, and enough grounding that the assistant can run on cheaper models. That's the mechanism — richer data in, cheaper computation out.

Will it actually ship?

That's the open question. The internal document reportedly says the project "depends on discovering how brands organize their products on their own websites," which could mean Amazon crawling other retailers' sites at scale. Reuters notes it's unclear "if or when this plan may be implemented."

Amazon pushed back on the reporting. In a statement to Reuters, the company said: "Draft, outdated documents often paint an incomplete picture and the existence of a brainstorming document such as this one doesn't necessarily reflect our current plans."

What's at stake?

For competitors, the stakes are structural. Google and Reddit currently capture the product-research phase of shopping — the questions, comparisons, and skepticism that precede a purchase. If Amazon absorbs that phase into its own product pages, it controls both research and checkout. For retailers whose products get indexed, the deal is murkier: Amazon's traffic could rise even as their direct traffic falls.

The 2027 date in the document suggests Amazon sees this as a multi-year build, not a near-term launch. But the direction is clear — the company that already indexes much of what's for sale online wants the rest of it, and wants AI agents reading from its copy.