Skip to content
MustHave.ai Musthave.ai
AI Startups & Business

Sierra releases Poppy 1.0. What Shopify merchants can check

Sierra says Shopify now publishes Poppy discovery files. Merchants can check discovery, product data and shopper authorization before checkout.

Sierra releases Poppy 1.0. What Shopify merchants can check
A discovery file tells a personal agent where to begin with a store.

Start with one product whose price and availability you know.

Sierra released version 1.0 of Personal Agent Protocol, which it calls Poppy, on November 13. It gives a personal AI agent a way to discover how a business wants to handle requests and which actions it exposes. Sierra says Shopify publishes the discovery file for every store and that Meta’s Muse looks for it on sites it visits. Those are Sierra’s rollout claims. Sierra’s release announcement

What the protocol does for a shop

A business publishes a discovery document that points agents to the capabilities it offers. The commerce extension uses Universal Commerce Protocol for shopping workflows. A store can use the discovery step to describe where to start, then rely on the applicable identity and authorization steps when the shopper asks for an action that needs their permission. Personal Agent Protocol documentation

For a merchant, the first check is what a visitor’s agent can actually retrieve. A discovery document shows that the domain can advertise capabilities. It gives no direct count of shoppers, product clicks, or completed orders. Each of those requires evidence from the relevant analytics or order system.

Test discovery and authorization before placing an order

First, check the domain you use for customers. Find the discovery document using the protocol specification and confirm that it describes the store you expect. If your setup has a separate shop subdomain or staging site, record which domain you checked.

Next, choose one product with a known price and stock state. Ask a compatible agent to find it. Compare currency, price, availability, and the product URL against the store. Then ask about a feature that requires signing in, such as your own order history, using a test account. Check that private information stays behind authentication.

Finally, test what happens when an action requires consent. You can stop at the permission or checkout step. Log where the handoff happened and what the customer was asked to confirm. That record makes it easier to tell which part failed if a later session reaches the wrong product or cannot continue.

The company implementation kit includes a validator that starts real test sessions and sends messages to the company agent. The kit recommends a test environment. Merchant coverage, real shopping traffic and production checkout should each be verified separately before reporting a result.

Abdessalam Alaoui

I'm Abdessalam Alaoui, founder of Musthave.AI. I write practical guides to AI models, tools, generative media.

Join the discussion

Your email address will not be published. Required fields are marked *

Keep reading

All news