Agentic Commerce
/eɪˈdʒɛntɪk ˈkɒmɜːrs/
Also known as: agent commerce, agentic shopping, AI shopping agents, selling to agents
What is Agentic Commerce?
Agentic commerce is buying and selling where an AI agent acts for the customer: it searches, compares, fills a cart, requests a quote or completes a payment within limits the person sets. The business on the other side is no longer selling only to a human looking at a page, but also to software reading it.
The Ladder of Trust
Greg Isenberg describes three rungs of agent shopping:
- Research: Find options, compare them, report back. Most businesses accept this because it is close to search.
- Prepare: Put items in the cart, hold seats, get a quote.
- Pay: Spend the customer's money. This is where most of the conflict and risk sit.
A typical instruction gives the agent a goal, rules and a limit, for example "find the right filter, compare the delivered price, and check with me before you buy anything over $50."
What Agents Read
Agents skip banners, videos and pop-ups. They read structured facts: total price, delivery date, compatibility, return terms. Information buried in a PDF or shown only on the last checkout screen is easy for them to miss. Standards such as UCP and trusted-agent programs from payment networks aim to give agents a reliable way in instead of scraping.
Why It Matters
Agents are disciplined shoppers. They compare suppliers in seconds, notice hidden fees and do not impulse buy, so they weaken revenue from ads and upsells. Businesses that are easy for agents to understand and buy from can win customers that competitors never see, which is why the war on agents is as much about business models as security.