When AI agents do the buying

An increasing share of purchase research is performed by a machine acting for a person. That machine reads the same web as a human and processes it differently.

What the intermediary reads

A business whose prices are only in a downloadable PDF, whose specification is in an image, and whose service area is implied rather than stated, is difficult to include in a shortlist.

What to do about it

Publish the facts as text. Prices, ranges, specifications, service areas, lead times, opening hours. Not as images, not only as downloads.

Use structured markup. Product, service, organisation, opening hours, review and price schemas make the information unambiguous rather than requiring inference.

Be consistent across sources. A machine comparing your website, your map listing and a directory entry and finding three different phone numbers will discount all three.

State what you do not do. Exclusions and limits help a machine correctly exclude you from irrelevant shortlists, which improves the quality of the enquiries you do receive.

The question of agent traffic

Requests from these systems appear in server logs and mostly do not appear in browser-based analytics, because they do not execute page scripts.

That produces a growing measurement gap: traffic that influences purchases and is invisible in your reporting. Server-side log analysis is the only way to see it.

What does not change

The human still makes the decision. Trust, reputation, reviews, the quality of the conversation once contact is made, and whether you answer the phone all remain decisive.

The machine narrows the field. Everything that has always mattered still decides the outcome within it.

The near-term practical position

Do not build for autonomous purchasing, which remains narrow. Do build for machine-readable research, which is already routine.

The overlap between the two projects is nearly complete, which means the work is worth doing regardless of how quickly the more autonomous scenario arrives.

The risk to watch

If intermediaries become the primary route to shortlisting, they acquire the position that search engines held, with the same consequences: a gatekeeper, an opaque ranking, and a market for placement within it.

Businesses that build a direct relationship with customers, so a list, a reputation, a reason to be asked for by name, are the least exposed to that shift, as they were to the last one.