How to become Agentic Commerce compliant with feed® PIM
AI-powered commerce is changing the way products are discovered, reviewed, and purchased. The Agentic Commerce Protocol (ACP) is an open standard that will allow customers to purchase products directly in ChatGPT via Instant Checkout.
When AI agents are to be able to find, understand and present products and connect them to the right online store - structured, complete and reliable product data is required.
This makes product data more important than ever, and it is in feed® PIM that the foundation is built. Without a solid product data engine at the bottom, you will be in danger of being prioritized.
The crucial principle for this:
AI chooses the products it understands best.
This is the core of why feed® PIM is one of the most important tools for making customers ACP compliant.
Understanding ACP's basic model: product data vs. trade data
ACP distinguishes very clearly between two data types:
Product data (feed® PIM):
GTIN/MPN/brand
Descriptions
Images and rights
Specifications
Compatibility
Categorization and metadata
Trade data (online store):
Price
Stock status
Shipping, tax and currency
Availability
Checkout, Payment and Merchant-of-Record
feed® PIM covers everything that belongs to the product data domain , while the online store takes care of the rest. This division of labor is crucial to becoming ACP compliant.
Build required ACP fields in feed® PIM
ACP does not require one rigid data structure, but a sufficiently complete and semantically clear model, and feed® PIM allows you to create all the fields you need:
Mandatory ACP-relevant fields:
Product name and short/long description
Brand/manufacturer
GTIN/EAN/UPC
MPN
Images with ownership information and license metadata
Main category (e.g. Google Product Category)
Highly recommended for AI agents:
Structured specifications
Feature bullets
Dimensions, weight, materials
Variant attributes
Accessories/compatibility
Packaging information
The more structured and complete this is in feed® PIM, the higher the chance that AI agents will prioritize and display the product correctly.
Ensuring data quality and ownership
ACP places great emphasis on trust in the data source . Therefore, feed® PIM should contain:
Image_rights / content_ownership
Supplier/brand relationship
Original_description_source
Data_source_id
This allows AI to identify you as the “source of truth,” and it improves both ranking and visibility in generative searches.
Ensure good, complete and structured product data
To deliver an ACP-compliant feed, your product data must be clear, consistent, and machine-readable. This means ensuring:
complete and correct product descriptions
unique values and standardized units
Complete and precise categorization
structured specifications and GTIN/MPN identifiers
When the product data in feed® PIM is well modeled and quality assured, the integration and online store can easily combine this with their own trade data (price, inventory, shipping) and generate an ACP-compliant feed without the need for heavy integrations or new systems.
Summary
AI trading is about clarity, quality and structure.
The products that AI understands best are the products AI chooses.
feed® PIM gives you the tools, flexibility and control you need to build ACP-friendly product data - and online stores are working to use this data to become fully-fledged Agentic Commerce players.
For those of you who want to learn more about how to get started with Agentic Commerce and ChatGPT Instant Checkout, we recommend visiting the official merchants page:
https://chatgpt.com/nb-NO/merchants/
Here is an overview of how trade data, product data, and checkout are connected in ChatGPT – and what is required to become part of AI-driven trade.
At the same time, it is important to remember that ChatGPT is not alone. Google Gemini, Perplexity and several other players are now building their own versions of AI-based shopping and agent-driven commerce. Good, structured and machine-readable product data in feed® PIM is therefore one of the most important things you do to become visible and relevant across all future AI channels.
