Agentic commerce

Agentic commerce is commerce in which the buyer is an autonomous software agent acting for a person — finding the product, choosing the variant, adding it to the cart and reaching checkout without a human reading the page at each step.

The buyer stops being an error detector

Every storefront has been quietly relying on a capability nobody specified: the buyer checks the work. A person scanning a cart notices the wrong size, the wrong colour, the quantity that reset. They correct it in a second and think nothing of it.

Remove that and a class of defect that was previously self-correcting becomes an order. The defect was always there. What changes is that it now reaches the end.

Agent-ready is not commerce-ready

Agent-ready describes a store an agent can operate: readable markup, working controls, published structured data, perhaps a machine-readable commerce interface.

Commerce-ready describes a store where operating it produces the correct commercial result.

The first is about access; the second is about outcome. Investment currently concentrates on the first, because it is visible, tractable and easy to demonstrate. The second is the one a buyer experiences.

Where journeys go wrong

From our own inspections, the failures that matter cluster in a few places:

Variant substitution at add-to-cart. The requested variant is unavailable and a default is silently used.
Option controls that display but do not apply. The label updates; the underlying variant does not.
Sibling SKUs. A near-identically named product is matched instead of the requested one.
Cart and channel disagreement. The storefront and the merchant's own cart API report different items.
Quantity drift. The requested quantity is not what the cart holds.

These are the shapes we have observed, described as shapes. We publish no failure rate: we have not run a study designed to support one, and a percentage without that design would be a number pretending to be evidence.

What a merchant can do about it

Make the store operable by a program — stable controls, honest availability, readable markup.
Verify that operating it yields the requested item, not merely a completed journey.
Check that any machine-readable channel agrees with the storefront.
Re-verify after theme changes, app installs and platform upgrades. See commerce regression testing.

Step two is the one usually skipped, because completion is easy to measure and correctness is not.

Next steps

Check your commerce journey — request an inspection of a store you are authorised to test.
CRAB Q&A — fifty questions on commerce execution assurance.
The Commerce Execution Method — the method itself, as a prepublication draft.
Commerce Execution in the Age of AI Agents — the whitepaper.

CRAB is built by VSN.