UCP, AI and Agentic Commerce
What's happening, and what's coming up next
Quick bonus post outside our usual cadence - this news is too fresh to sit on.
Shopify and Google just launched the Universal Commerce Protocol - an open standard for AI agents to discover products, negotiate checkout, and complete transactions.
This is the infrastructure we’ve been writing about. The shift from websites-for-humans to structured data that agents can query, understand, and act on. NLWeb, Schema.org, MCP - these weren’t just tech experiments. They were signposts. And now the biggest commerce platform on the planet has formalised it into an open protocol.
UCP handles the hard stuff: loyalty programs, dynamic pricing, subscriptions, embedded checkout. Agents and merchants declare capabilities, the protocol negotiates the difference. No integration meetings required.
But here’s the thing most brands will miss: the protocol is only as good as your product data.
UCP can’t magic clean attributes out of a tangled PIM. It can’t surface which products suit a three-year-old if that information lives nowhere in your data model. It can’t make sense of “footwear” vs “ankle boots” vs “shoes” scattered inconsistently across your catalogue.
And in the agentic web, messy data doesn’t just slow you down. It makes you invisible.
The brands that win this next era of commerce won’t just be the ones who implement UCP first. They’ll be the ones whose product data is already structured, semantic, and machine-readable. The ones who understood early that product data isn’t just an operational asset - it’s becoming your interface to the world.
We’re working with some of the biggest retailers in ecommerce right now, helping them navigate exactly this shift. What we’re learning, we’re documenting here.
Watch this space.