Alipay launches an agentic commerce platform for merchants

What happened

At its AI Ecosystem Partner Conference in Hangzhou, Alipay launched what it describes as China's first full-stack agentic commerce platform, giving merchants the tooling to turn existing pages, listings and service workflows into agent-ready Skills and MCP services (Ant Group press release, August 2026).

Ant Group CEO Cyril Han said agentic commerce is set for rapid growth over the next 6 to 12 months, with AI agents becoming an interface between hundreds of millions of consumers and tens of millions of merchants (Yicai Global).

Why the merchant side is the news

Consumer-facing AI shopping has been shipping for over a year: assistants that search, compare and build carts. What has been missing is the other end of the pipe, where a business exposes its own catalogue, pricing and workflows as callable tools instead of hoping a crawler parses its pages correctly (The Paypers).

That is what makes this worth reading outside China. A payments company decided the durable layer is merchant infrastructure, not another chat interface.

MCP is becoming the default plug

The Model Context Protocol started as a way to give an assistant access to files and internal tools. In the space of a year it has become how commerce systems expose themselves: retailers as shopping connectors, ad platforms as reporting tools, and now a payments network as a merchant platform.

For an Amazon seller the practical read is simple. The interface that will matter for running the business is a conversation with an assistant that can call your systems, and the systems worth connecting are the ones holding money data: profit, fees, advertising spend, inventory.

What to take from it this quarter

1. Fix your product data before you worry about agents. Attributes, variation consistency and accurate pack sizes are the inputs any agent reads. None of that is speculative work.
2. Decide what an assistant is allowed to touch. Read-only access to reporting is a different risk profile from write access to pricing or inventory. Start with read.
3. Instrument AI-referred traffic. You cannot argue about the channel later without a baseline now.
4. Connect your own numbers first. The fastest payback from agentic tooling today is operational: answers about your business, not new demand.

The Nova angle

Nova exposes an Amazon seller's reconciled profit, advertising, search query, inventory and listing health data to an assistant over MCP, with read access and no data export. See Nova MCP for Amazon sellers or the wider AI agents hub.