CorgiMaps adds ARD and AI Catalog discovery protocols

Before an AI agent can buy from a store, it has to find it. CorgiMaps now supports two new agentic commerce protocols built for that first discovery step: ARD (Agentic Resource Discovery) and AI Catalog.

Discovery comes first

Checkout protocols like UCP and ACP define how an agent buys. Discovery protocols answer an earlier question: what's out there, and where?

ARD and AI Catalog both help a business publish their AI-ready resources, such as its MCP servers, its A2A agents, and more. Search indexers and AI platforms read those discovery surfaces, so when an agent goes looking for something that can help with a task, the brand's store is in the answer.

These protocols are about getting a brand discovered and recommended by AI platforms, not about what happens once an agent arrives. They're the kind of protocols other commerce platforms will lag in supporting, because they sit outside the checkout flow. That's exactly why we're supporting them now.

ARD

ARD lets a business publish a list of its agent-callable resources on its own domain, each with a short description and a few example requests it's good at. Discovery services index those lists and match them to what agents are asking for. ARD only handles discovery. Each resource is then used through its own protocol for catalog search and checkout, such as UCP, MCP or A2A.

Google announced ARD publicly in June. Google, Hugging Face, Microsoft, Amazon and Cisco sit on its oversight board, and its contributors include GitHub, Salesforce, Nvidia, ServiceNow and Snowflake. Google has promised native ARD support in Gemini.

ARD is still a proposal, but we've been tracking it for a few months and it has stabilized enough to support today. If anything changes when it reaches general availability, CorgiMaps will keep up, the same way it does for every protocol it supports.

AI Catalog

AI Catalog is another emerging discovery protocol that lists everything AI-related a domain offers, from MCP servers to A2A agents. It's maintained by a working group under the Linux Foundation, with a steering committee from Google, Microsoft, and Anthropic. AI Catalog is still a draft as well.

The two protocols are closely related. ARD builds on AI Catalog's data model.

What brands get

• ARD and AI Catalog support. Turn them on under Protocol Settings. There's nothing else to configure.
• A catalog on your agent domains. CorgiMaps publishes a catalog that lists your UCP, ACP and MCP servers, as well as your A2A agent.
• Your own example requests. Brands can provide their own example requests that go into your ARD entries, so discovery services know how you talk about your products and services.

This CorgiMaps release also adds:

• Listing on MCP registries. CorgiMaps can automatically register your store on MCP registries automatically. Registries are the directories agents search when they look for a tool.
• An MCP server card. Another way to help brands get discovered, CorgiMaps customer’s agent domains publish a card that agents and indexers read to find your MCP server and connect to it.

"Getting found is the first problem every brand has in AI shopping, and discovery protocols are how agents will solve it," said Jonathan Porter, co-founder and Head of Product & Engineering at CorgiMaps. "These new protocols have the companies building the AI platforms behind them, and the protocols are stable enough that we felt it’s time to start supporting them. Our customers are listed today, and if the specs move, we move with it, automatically."