Agent Commerce: How to Get Your Products Bought by AI Agents
By Mohamed Lotfi SassiPublished 2 min read
For twenty years, selling online meant one thing: win the click. Better ads, better landing pages, a faster checkout.
That's starting to change. More and more, the first visitor to compare your products won't be a person. It'll be an AI agent working for one, and it doesn't care how nice your homepage looks. It cares whether it can read your catalog and trust your data.
What analysts expect
Gartner forecasts that AI agents will intermediate $15 trillion in business purchases by 2028. Forrester predicts that in 2027, one in twenty US and UK consumers will replace traditional shopping with conversations, and that a third of commerce service providers will launch agentic commerce platforms.
How agent buying works
An agent receives a goal, such as "order 40 ergonomic chairs under $300, delivered by Friday". It then:
- Searches suppliers it can read and trust.
- Compares specifications, prices, stock and delivery.
- Negotiates within the limits it was given.
- Places the order, often asking a person to approve.
Suppliers that agents cannot read are simply skipped.
How to prepare
- Make your catalog machine-readable. Clear product data, prices and stock levels, available through an API or an MCP connector.
- Publish your rules. Shipping times, return policies and volume discounts should be explicit, not hidden in PDFs.
- Answer agents, not only people. A sales agent of your own can respond to buyer agents instantly, within the limits you set.
- Protect against agent fraud. Verify which agents are allowed to order, and set spending limits.
Where to start
Start with a connector that lets AI assistants read your catalog and stock. It is the same work that puts your product inside Claude and ChatGPT, and it prepares you for agent buyers at the same time.
Sources: Gartner forecast via MarketScale, Forrester Predictions 2027