Agentic commerce examples, traced from the request to the checkout
Four agentic commerce examples follow an AI agent from the shopper's request to the checkout and show where the purchase breaks. Two are audits of real Swiss shops, measured and anonymised, and two are constructed worked examples that show the mechanics. Each page says which one it is.
What does agentic commerce look like for a real shop today?
Today it mostly looks like an agent that finds a shop, reads its product pages and stops before buying, because the price or stock is not declared as data. Agentic commerce is shopping in which an AI agent searches, compares and buys on a person's behalf, as set out in what agentic commerce is. In our audit of 240 online shops in French-speaking Switzerland, 157 of the 211 read in full lost most agents at that product data stage (July 2026).
| Case | Kind | Sector | Where the purchase breaks | The fix |
|---|---|---|---|---|
| How ready are Swiss online shops for AI shopping agents? | Audit, 240 shops | Online retail, French-speaking Switzerland | Understand: 157 of 211 shops do not declare price, stock and identity as data | Product and Offer markup generated from the catalogue |
| A chocolatier AI agents could recommend but not buy from | Audit, one shop | Food and confectionery | Understand: named in 58% of Google AI Overviews tested, no structured price on 26 product pages | Product, Offer, Organization and rating markup |
| How an AI shopping agent compares gravel bikes under CHF 4,000 | Worked example | Sports retail | Understand: the cheaper shop loads price and stock by JavaScript and has no GTIN | Price, stock, delivery and GTIN as structured fields |
| How an AI purchasing agent reorders office supplies under a standing mandate | Worked example | B2B supplies | Buy: the supplier cannot verify the signed mandate | MCP stock and cart tools with AP2 mandate checks |
Which of these examples are measured and which are constructed?
The two audits use data our crawler measured on real shops in July 2026, and they name no shop. The gravel bike and office reorder cases are worked examples with invented shops and illustrative figures, built to show which fields and protocol steps an agent checks. For the industries behind each case, agentic commerce by industry lists the fields per sector, and the agentic commerce readiness check runs the same missions on your own shop.
-
Audit Online retail, French-speaking Switzerland
How ready are Swiss online shops for AI shopping agents?
Are Swiss online shops ready for AI agents? Our July 2026 audit of 240 shops in French-speaking Switzerland found none scoring an A and a median of 57.
-
Worked example Sports retail, Switzerland
How an AI shopping agent compares gravel bikes under CHF 4,000
How AI shopping agents compare products, in a worked example: the bike shop whose price, stock and delivery are structured fields beats a cheaper rival.
-
Worked example B2B supplies, DACH
How an AI purchasing agent reorders office supplies under a standing mandate
Worked example: a Zurich office's AI purchasing agent reorders paper and coffee under CHF 400, and what a supplier must expose, up to AP2 mandate checks.
-
Audit Food and confectionery, Switzerland
A chocolatier AI agents could recommend but not buy from
July 2026 audit: a chocolatier near Geneva was named in 58% of Google AI Overviews, yet none of 26 product pages declared a price an agent could read.
Have your own shop run through a mission
A Buyer Journey Check runs missions like these through ChatGPT, Perplexity, Claude and a test agent on your own categories, then hands your developers the fixes in the order that recovers the most sales.