Two industries. One readiness standard.
Retail and travel are the two categories where agents can already research, compare and complete a purchase. The systems are different. The six layers are the same.
Your catalog has to survive a machine.
AI-referred traffic to US retail sites more than doubled year over year in May 2026, according to Adobe Analytics. By October, Adobe found AI traffic generated 83% higher revenue per visit. The shopper arrives ready to buy. The question is whether the agent can read your product, price and stock.
What agents need
- A current product feed with price, stock, variants and images.
- Product, Offer and AggregateOffer schema on the page.
- Shipping, returns and delivery terms in structured form.
- A checkout that accepts agent payment tokens.
- An order confirmation the agent can read.
What usually breaks
- The feed is stale, incomplete or missing variants.
- Product pages lack a usable Product node.
- Shipping and returns are missing from the Offer.
- The firewall challenges verified agents.
- The checkout rejects the token, or the risk rules decline it.
What the scan found
An independent September 2026 scan of 35 stores. Small, hand-picked sample, not a random survey.
Published a UCP manifest.
Had a usable Product node.
Exposed shipping or returns in the Offer.
Showed a captcha to an identified agent.
The gap is data, not access. Most agents can get in. Few can work out what the product costs, what delivery involves or what the return terms are.
Who this is for
DTC brands, mid-market retailers and marketplaces with a direct checkout. The best fit has a product catalog, a feed and a team that can change the site.
What we fix
Feed and schema, verified agent access, checkout and payment tokens, shipping and returns data, order confirmation and measurement. One owner, one roadmap.
Your property has to be quotable and bookable.
Travel is the highest-value, most intermediated purchase an agent can make. Phocuswright finds 40% of US travelers would let an AI assistant book flights and hotels, against 28% in the UK. IATA finds 49% have used AI tools to plan or book travel. The gap between what an agent can plan and what it can book is the opportunity.
What agents need
- Property, room and rate data in a form a machine can read.
- Cancellation, deposit and check-in policies as structured data.
- Availability for specific dates and occupancy.
- A booking engine or API that can quote, hold and confirm.
- A payment token the checkout accepts.
- Confirmation, modification and cancellation the agent can read.
What usually breaks
- Rates sit behind a booking form.
- Policies sit in a PDF or an image.
- The booking engine has no API for agents.
- The payment token is rejected, or the risk rules decline it.
- No one can see the agent traffic in analytics.
Why travel is different
The purchase is high value, multi-component and already intermediated. An agent can plan a whole itinerary, but it cannot confirm the room, the rate and the cancellation terms without a bookable component.
The suppliers that expose clean data and a real booking action become the components the agent uses. The ones that do not get skipped, or defaulted to an OTA.
Sources: Phocuswright, The AI Surge, March 2026; IATA survey reported by Business Traveller, November 2026.
Who this is for
Independent hotels, DMCs and experience operators with a direct booking engine. The best fit has a named owner, access to the engine and a reason to grow direct bookings.
What we fix
Property and rate data, policies, verified agent access, booking tools, payment tokens, cancellation and measurement. We do not replatform for the sake of it.
The teardown tells you.
Send the site. I will tell you what an agent can do with it, whether you sell products or rooms, and what to fix first.