Amazon reportedly wants to build a more comprehensive picture of products sold across the internet. For businesses, the bigger lesson is that AI cannot make dependable decisions from information it doesn’t understand.
Imagine a distributor asking an AI assistant to find a replacement component for a customer.
The assistant identifies a product, compares specifications, checks availability, and prepares a quotation. The process takes minutes instead of the usual back-and-forth between Sales and Operations.
But there’s a problem. The product description is outdated, the inventory record reflects yesterday’s stock, and the customer has a negotiated price that the assistant cannot see.
The AI responds confidently. The quotation is wrong.
This example illustrates a challenge businesses face as artificial intelligence becomes more deeply connected to product catalogs, customer records, and operational systems.
AI can retrieve and interpret information quickly. It cannot automatically make that information accurate.
Amazon’s Reported World Product Graph Puts Data Quality in Focus
On November 13, 2026, Reuters reported that Amazon is developing a proposed initiative called the World Product Graph, intended to assemble extensive information about products available across the internet, including products not sold directly through Amazon.
According to the report, internal planning documents describe an effort to improve product discovery and support Amazon’s AI shopping capabilities. The documents also acknowledge problems with incomplete or inaccurate AI-generated product answers and the cost of operating AI shopping systems.
The initiative could expand Amazon’s ability to answer product questions and compete with external AI shopping assistants. However, Reuters noted that the plans may change, and the reported initiative should not be treated as a completed product. Source: Reuters, November 13, 2026.
The business lesson extends beyond online shopping.
As AI systems become interfaces for finding products, preparing quotations, and answering customer questions, the underlying data becomes increasingly important.
What Is AI Product Data Management?
AI product data management is the practice of organizing, maintaining, and governing product information so AI-enabled applications can use it reliably.
That information may include product names, identifiers, descriptions, specifications, pricing, availability, compatibility, and customer-specific terms.
In many businesses, these details are distributed across multiple systems.
A CRM might contain customer relationships and negotiated pricing. An ERP system might manage inventory and fulfillment. A product catalog may hold technical specifications. Employees may maintain additional information in spreadsheets or shared documents.
Each system serves a purpose, but the information must remain consistent.
If an AI assistant retrieves the wrong product specification or uses an outdated price, the speed of its response becomes irrelevant.
The challenge is not simply making more information available. It is ensuring the right information is available for the right decision.
Why CRM and ERP Integration Matters
Consider a manufacturer selling products through distributors.
A distributor requests a quotation for a custom order. Sales maintains the customer relationship in Microsoft Dynamics 365 or HubSpot, while an ERP system manages product information, inventory, and production planning.
Preparing an accurate quotation may require information from both systems.
The CRM identifies the customer and opportunity. The ERP provides product details and operational information. Pricing rules may depend on customer agreements, order quantities, and other conditions.
An AI assistant could help coordinate these activities, but only if the information is connected and the business rules are clear.
Simply giving an AI tool access to both systems does not resolve inconsistencies between them.
For example, Sales may consider a product available when it can be manufactured within the customer’s requested timeframe. Operations may define availability as inventory physically on hand.
Neither definition is necessarily wrong. They answer different questions.
The organization must establish which definition applies to the workflow before automating the decision.
AI Search Makes Inconsistent Information More Visible
When employees work manually, they often recognize exceptions through experience.
A sales representative may know that a particular product has been discontinued. An operations employee may remember that a supplier recently changed a specification. A customer service representative may recognize that a customer uses a special product configuration.
These details may not be reflected consistently in business systems.
An AI assistant does not necessarily possess that informal knowledge.
If it retrieves incomplete records, it may produce an answer that appears reasonable but fails to reflect the actual business situation.
This creates an opportunity to improve how organizations manage information.
Rather than treating data cleanup as a one-time technical project, businesses can establish ongoing responsibility for product records, customer agreements, and operational updates.
The objective is to make reliable information part of everyday work.
Five Steps to Prepare Product Data for AI
Organizations do not need to rebuild every system before exploring AI. They can begin with the information supporting one important business process.
- Identify the decisions AI will support. Determine whether the goal is product discovery, quotation preparation, customer service, inventory inquiries, or another activity. Different use cases require different information.
- Establish authoritative sources. Decide which system owns each category of information. For example, ERP may be authoritative for inventory while CRM maintains customer-specific relationships and agreements.
- Standardize definitions and identifiers. Ensure product numbers, units of measurement, statuses, and customer references are interpreted consistently across systems.
- Define ownership and update responsibilities. Assign people or teams to maintain important records. Establish procedures for handling discontinued products, revised specifications, pricing changes, and exceptions.
- Test AI against realistic business scenarios. Evaluate whether the system handles unavailable products, conflicting records, customer-specific pricing, and incomplete information. Measure accuracy and the amount of employee correction required.
These steps help organizations improve information reliability whether they ultimately use AI, traditional automation, or both.
Better Data Requires Better Collaboration
Product information rarely belongs to one department.
Sales understands customer expectations and commercial agreements. Operations understands production capabilities. Finance maintains pricing and margin requirements. IT manages the systems and integrations connecting the information.
Each department sees a different part of the business.
At DKomplex, our people-first IT+® approach emphasizes bringing these perspectives together before configuring technology.
Our collaborative problem-solving workshops help organizations examine assumptions, clarify responsibilities, and establish shared definitions.
We call this developing coherence: creating a common understanding of how people, processes, and information work together.
For example, a request to automate product quotations may reveal that the real obstacle is not the time required to create a document. It may be disagreement about pricing authority, product availability, or when an opportunity is ready to advance.
Resolving those questions can improve the process even before AI is introduced.
The Competitive Advantage Is Trustworthy Information
Amazon’s reported initiative illustrates how valuable comprehensive product information may become as AI changes how customers research and purchase products.
For manufacturers, distributors, and other businesses, the practical opportunity is to make their own information more dependable and accessible.
That does not necessarily require creating a massive database or purchasing another application.
It may begin with improving existing CRM records, clarifying product definitions, connecting systems, and establishing ownership of important information.
AI can make it easier to retrieve information and coordinate work. But organizations must still determine what information means, who is responsible for maintaining it, and how it should be used.
The businesses that benefit most from AI may not be those with the most data. They may be those whose people and systems can agree on what their data means.
Make Your Business Data More Useful With DKomplex
DKomplex helps organizations improve customer relationships and business operations through Microsoft Dynamics 365, HubSpot, Power Platform, CRM consulting, systems integration, and collaborative problem-solving.
Our IT+® approach begins by understanding the business problem, involving the people responsible for the work, and designing technology around practical outcomes.
Explore our Problem Solving as a Service workshops or connect with DKomplex to discuss how better information and connected systems can support your organization’s next technology initiative.
DKomplex | Complex problems made simple.