TRV-2026-2902Version 1 · Published

Saved · see the current record

Why this version exists

Certified into the record

The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-2902
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-14T00:02:36.657772Z
status: published
lens: p_space
sector: business
headline: Amazon’s plan to own the world’s product data and fight AI shopping agents
dek: Amazon is developing World Product Graph to compile global product data, aiming to challenge Google search dominance, rival Walmart, and improve Rufus AI.

Amazon has a secret plan to ensure shoppers never need to leave its site: own the world’s product information.
Called World Product Graph, the initiative seeks to hoover up data on virtually every item for sale on Earth, whether Amazon stocks it or n
gain_title: (none)
problem_title: An internal Amazon document reviewed by Reuters says many answers from its Rufus shopping AI are factually wrong or incomplete and shoppers cannot spot the errors, though Amazon says the assistant is roughly 97% accurate.
trace_subject: (none)
gain_reading: (none)
gain_evidence: (none)
problem_reading: An internal Amazon document reviewed by Reuters says many answers from its Rufus shopping AI are factually wrong or incomplete and shoppers cannot spot the errors, though Amazon says the assistant is roughly 97% accurate.
problem_evidence: Many Rufus answers are factually wrong or incomplete, but customers lack expertise to detect these errors
quick_read: An internal Amazon document reviewed by Reuters describes a project called World Product Graph, which would collect data on virtually every item for sale, whether or not Amazon stocks it. The document says the company's own AI search agent, Rufus, often provides inaccurate information.

The plan is also a defence against outside AI shopping agents that could send customers to sites with more complete product data. Amazon has blocked Meta's Muse agent and sued Perplexity over similar access, and CEO Andy Jassy has played down the agents' impact so far.

Amazon says the document is a draft brainstorming memo that does not necessarily reflect current plans. Reuters could not determine whether or when the project will launch, and the memo carries an aspirational release date of October 1, 2027.
limitation: Reuters could not determine if or when the plan will be implemented, and Amazon says the draft document is outdated and does not necessarily reflect current plans, so the Rufus accuracy and usage figures are disputed.
tag: Evidence-backed problem
key_points: The document says Rufus answers are often factually wrong or incomplete and customers lack the expertise to detect the errors. | Users run about 83 million Rufus searches a week, and the document calls running them expensive; more data would let Amazon use cheaper models. Amazon says those usage figures are significantly outdated. | Amazon says its data shows Alexa for Shopping, as Rufus has been called since May, is roughly 97% accurate globally. | A 2025 Morgan Stanley report found 16% of online shoppers start product research on Amazon versus 57% on Google; when the product is already chosen, 41% of searches start on Amazon, 23% on Google and 18% on Walmart. | Amazon has signed Coca-Cola and Bayer, among others, to contribute product data, and held a CEO-level summit for a trial with Procter & Gamble. | Meta's Muse agent has more than 10 million users in the US and Canada since launching in September; Amazon blocked it from its retail site that month. | Amazon sued Perplexity last year to stop similar access, and that case is still ongoing.
rundown: The document is a PR/FAQ, the roughly six-page internal narrative memo Amazon uses to make the case for a new product. It says the plan depends on discovering how brands organise products on their own sites, which would involve some brands and retailers feeding data to Amazon and Amazon crawling other retailers' websites. Reuters could not determine how the crawling would work or whether permission would always be sought.

The project sits in Amazon's Selection & Catalog Systems unit under vice president Robert Tekiela. Staff are being shuffled internally to start it, according to one person familiar, but the people said it could be modified or aborted over financial or other concerns. With fuller data, Amazon could tell shoppers which vacuum cleaner or tablet is the latest model and steer them by personal criteria such as shopping history.

On agents, Jassy said on the third-quarter earnings call that few customers shop this way yet and that Amazon's advertising business has been minimally affected. People familiar said executives are nonetheless wary that agents will route traffic to sites with more complete data. JPMorgan said Muse could become the most widely used consumer AI application since ChatGPT, and Elon Musk wrote on X that perhaps he should start a competitor to Amazon after a post about Grok being blocked.

Employees describe World Product Graph as an expansion of existing Buy for Me and Shop Direct services, which list products sold on other sites and let customers buy them through Amazon as an intermediary. Jassy said Alexa for Shopping has been used by over 400 million people in the past year.
sources:
- journalism | CNBC TV18 | https://www.cnbctv18.com/technology/amazons-plan-to-own-the-worlds-product-data-and-fight-ai-shopping-agents-20033618.htm | 2026-11-13
prev: 0000000000000000000000000000000000000000000000000000000000000000
SHA-256 hash
f7b9e6d9ca394257ba50b9fec247461cd136579d5a890e1d5dad74476f5bd94c
Previous version’s hash
0000000000000000000000000000000000000000000000000000000000000000
Check this record hasn’t been altered
How to check it yourself, without trusting this page

Download any version from /api/record/TRV-2026-2902 and hash its canonical field yourself, for example with shasum -a 256. You should get its content_hash. Each version’s text also ends with prev:and the hash of the version before it (version 1 points to 64 zeros). If anyone had changed a single character of any version after it was saved, those hashes wouldn’t line up.