What Is Agentic Commerce? Definition and Example

Agentic commerce is shopping in which an AI agent discovers, compares and sometimes buys products on a person's behalf. Instead of browsing many sites, the buyer describes what they want, and the agent handles research and, increasingly, checkout. It depends on machine-readable product data, trusted information and secure payment flows.

How does it work?

A typical flow:

1. The buyer states a need: "Find a waterproof hiking jacket under €200 that arrives by Friday."
2. The agent searches product data, sites and reviews.
3. It shortlists and explains options.
4. With the buyer's approval, it completes checkout, through a merchant integration or by browsing.

What infrastructure exists?

OpenAI documents an Agentic Commerce Protocol with specifications for product feeds, agentic checkout and delegated payment, so merchants can share structured product data and support purchases in ChatGPT. Google uses Merchant Center product data across its shopping experiences (Google). Details and availability change, so check current documentation.

What does it mean for merchants?

What are the risks?

• Wrong data at scale: an outdated price in a feed can be repeated to many buyers.
• Less brand contact: the buyer may never visit your site.
• Unclear attribution: agent-assisted sales can be hard to track.

Example

Illustrative. An outdoor retailer finds that agent shortlists for "lightweight tent for two" favour competitors. Its feed lacks weight and packed-size attributes. After adding them, the retailer monitors whether its products appear more often.

Related

• AI agent
• AI shopping
• MCP