Service pillar

Universal Commerce Protocol: getting WooCommerce ready for AI agents

Agents read the Schema.org layer for browsing and the MCP layer for transactional intent. WordPress remains the canonical source. AI agent (Claude, ChatGPT, browser) reads Schema.org Product + AggregateOffer for browsing and exchanges transactional intent via MCP server (typed intent). WordPress / WooCommerce origin holds the Canonical catalogue + orders. Webhook on price / stock change. AI agent (Claude, ChatGPT, browser) Schema.org Product + AggregateOffer machine-readable read MCP server (typed intent) transactional intent WordPress / WooCommerce origin Canonical catalogue + orders Webhook on price / stock change
Agents read the Schema.org layer for browsing and the MCP layer for transactional intent. WordPress remains the canonical source.

What I ship

  • Audit of the existing WooCommerce or WordPress catalogue for agent-readability, including the current state of REST endpoints at /wp-json/wc/v3/ and any structured data already present.
  • Schema.org enrichment per product, covering Product, Offer and AggregateOffer with hasMerchantReturnPolicy, ItemAvailability and PriceSpecification. The output is JSON-LD on the page and validated against schema.org.
  • Optional Model Context Protocol server exposing catalogue and order intent endpoints to authenticated agent clients, sitting alongside the WordPress REST API rather than replacing it.
  • Test harness wired to the Anthropic Claude API that exercises the agent flow end to end, from product lookup through order intent, with recorded transcripts as the regression baseline.
  • Observability that separates agent traffic from human traffic at the analytics layer, so attribution, conversion and cost can be reported per audience type.

Why this is not just another schema.org tweak

AI agents are starting to behave like buyers. Browser-resident assistants and agent surfaces inside Claude and ChatGPT increasingly act on behalf of users: filtering, comparing, eventually committing to a transaction. The interface they need is not the visual page; it is a machine-readable surface for the catalogue and the transactional flow on top of it.

Schema.org Product, Offer and AggregateOffer cover the description side. They tell an agent what a product is, what it costs, whether it is in stock and what return policy applies. They do not describe how an agent should authenticate, how it should signal intent to buy, or how the store should respond when many agents query inventory simultaneously. That is the gap converging standards are addressing.

Anthropic's Model Context Protocol, first published in November 2024 as an open specification, is one of the moves toward agent-readable interfaces. Automattic's position post on 2026-04-21, framing WordPress as a candidate operating system for agent-readable websites, points in the same direction from the CMS side. The practical conclusion: a catalogue legible only to humans is being repriced by the market.

Who this is for

  • WooCommerce stores where agent-driven traffic is already showing up in logs and the catalogue is not built to answer it.
  • B2B catalogues with quote workflows, regional pricing and subscription products, where a typed agent surface beats fragile screen-scraping.
  • Content sites with affiliate commerce, where the page is the catalogue and agent-readable Offers determine whether the site is cited or skipped.
  • Multi-region brands with one CMS feeding many storefronts, who need a single source of truth that agents can resolve consistently across locales.

Engagement model

Senior B2B contracts on EU jurisdiction. Discovery, scoping, fixed-scope or time-and-materials engagements. Scope and pricing are individual per project; I do not publish standard rate cards because the audit step usually changes the cost of the build.

It can be given tasks that take hours, days, or weeks to complete, and then goes off and does those tasks autonomously.
Dario Amodei , CEO of Anthropic , Machines of Loving Grace , 2024-10-01 , source

Frequently asked questions

What is the Universal Commerce Protocol?

Universal Commerce Protocol is shorthand for the cluster of work needed so that an AI agent acting on behalf of a buyer can read your catalogue, understand pricing and availability, and signal order intent in a structured way. It combines Schema.org Product, Offer and AggregateOffer markup with an optional Model Context Protocol (MCP) server that exposes a typed surface to authenticated agent clients.

How does this differ from Schema.org Product markup?

Schema.org Product, Offer and AggregateOffer describe the catalogue. They are necessary but not sufficient. They do not describe a transactional flow: how an agent should ask for stock at scale, how it should signal intent to buy, or how authentication and idempotency are handled. Anthropic's open Model Context Protocol, first published 2024-11, is one of the standards converging on that gap. I layer the two so structured data and a typed agent surface stay in sync.

Is MCP required?

No. Many catalogues benefit immediately from disciplined Schema.org enrichment alone, especially adding hasMerchantReturnPolicy, ItemAvailability and PriceSpecification. I add an MCP server when the client wants a typed, authenticated surface for agent clients beyond what crawlable JSON-LD can express, or when WooCommerce custom flows (subscriptions, B2B quotes, regional pricing) need a deterministic interface for agents.

Will this work on a non-WooCommerce WordPress site?

Yes. WordPress 6.7+ exposes content through the REST API at /wp-json/, and WooCommerce extends that at /wp-json/wc/v3/. For non-WooCommerce sites with affiliate commerce, custom post types or a headless setup, I map your existing catalogue model into Schema.org and, where useful, an MCP surface. The pattern is the same; the source of truth changes.

How do I measure ROI on agent-readability?

I split observability between agent traffic and human traffic on the analytics layer, so you can see which sessions originate from AI assistants and how their conversion behaviour compares. Pricing on the engagement is individual; what I track on outcome is agent-attributed revenue, cost-per-agent-session and the share of catalogue surface that is actually legible to agent clients.

Internal AI governance before public agent surfaces

Agent-readable commerce needs governance before exposure. The same discipline used for internal AI content operations applies to catalogue tools, MCP surfaces, and agent observability.

Read the anonymised AI content-ops case study

Cluster reading

The supporting articles in this cluster, by topic.

Agent-ready commerce and AI integrations

Architecture and edge

Reference

Start a B2B engagement

Tell me the scope and timeline, I reply within one working day.

Contact me