What agent-readiness is
Agentic commerce changes who makes the purchase decision. Where the shopper used to open a site and
search, in the agent model an AI does that — reading data directly through APIs, protocols and
machine-readable formats.
Agent-readiness is how prepared a store is for that interaction. A beautifully designed site with no
structured catalogue and no API can be entirely invisible to agents.
Four layers of readiness
1. The catalogue
Every product has to be machine-readable: complete attributes, accurate descriptions, current prices
and availability. An agent cannot infer a specification from a photograph — it needs explicit text.
2. Discovery
The agent has to find the store and understand what it can do:
- Schema.org Product markup on product pages
- A structured product feed
- An llms.txt describing the available capabilities
3. Order protocols
Standard APIs for completing a purchase with no redirect to the site:
| Protocol | Ecosystem |
|---|---|
| ACP (OpenAI and Stripe) | ChatGPT, Shopify agents |
| AP2 | The Google A2A ecosystem |
| Universal proposals | Multi-vendor compatibility |
4. The payment layer
Support for a shared payment token and compatibility with payment mandates. For most retailers this
is configured through the payment provider rather than built from scratch.
Agent-readiness as a competitive edge
Important: in 2025–2026 agent-readiness is an early signal, not a mass requirement. Stores that
implement protocol support first gain the advantage at the moment agent traffic reaches meaningful
volume — as the early adopters of mobile SEO did a decade earlier.
Starting small is reasonable: audit catalogue quality → add Schema.org markup → publish a product
feed → implement the first protocol relevant to your audience.