The practical definition
For merchants, agentic commerce is not just a chatbot checkout. It is a data-readiness problem across product feeds, policy pages, structured data, crawler access, protocol monitoring, and payment handoff.
What Is Agentic Commerce?
Agentic commerce is the shift from users manually browsing stores to AI agents helping discover, compare, select, and sometimes initiate shopping workflows.
What changes for ecommerce teams
- Products need complete, consistent, machine-readable data.
- Shipping, returns, price, availability, and trust signals need to be clear.
- Teams need to decide how AI crawlers and shopping agents can access public pages.
- Payments and checkout owners need to track agent-mediated handoff rules.
Example: a delivery deadline changes the purchase
Imagine a shopper asks for a waterproof jacket under $150 that arrives before Friday. An assistant can shortlist products, compare sizes and delivery promises, and direct the shopper to a checkout. In a supported integration, it may also prepare a cart or initiate an order after the required authorization.
The merchant's risk is a mismatch between those steps. A product may look suitable in a feed while the chosen size is out of stock, or the delivered total may exceed the shopper's budget. Our suggested test is to change one condition before payment and check that the user sees the revised offer and can decline it.
Discovery, checkout and permission are separate
A crawler reading a product page does not prove that a shopper saw it. An AI referral does not prove an agent placed an order. A protocol implementation does not prove channel eligibility. Record each step separately so a team can distinguish a catalog improvement from a completed transaction.
ACP and UCP give integration teams concrete interfaces to investigate. The practical starting point for a store is still a testable product offer, clear policies and a supported payment route. See our Shopify, WooCommerce and B2B evidence guide for examples.
What to do first
Start with readiness, not speculation. Run a catalog and policy audit, assign an owner for protocol changes, and keep a source-backed changelog of platform moves.
Run readiness check