ResourcesGuide
On-site AI assistant or visiting shopping agent: which job are you solving?
Decide whether to fix Shopify product data, build a store assistant, or test tools for agents shoppers bring to your site.
3 min read
“We need AI shopping” can mean three different customer experiences. A shopper asks your store’s assistant a question. A shopper brings an agent in their browser to your storefront. Or a shopper discovers a product in an outside AI channel before visiting you. The data, controls and evidence differ, so choose the experience before choosing an app.
Start with the shopper’s location
Scroll sideways for columns out of view. Keyboard: focus the table and use the Left and Right arrow keys.
| Desired experience | First investment to inspect | How to test it |
|---|---|---|
| A chat experience you operate on your site | Merchant-operated assistant, its answer sources, handoff and service ownership | Ask it a real presales question; inspect the source and escalation behavior. |
| A shopper’s compatible agent acts on your existing storefront | Native Shopify WebMCP tools and the product facts they return | In a supported browser-agent setup, ask one exact product question and inspect actual tool responses. |
| Product discovery in external AI channels | Shopify Catalog eligibility, product data, mapping and channel controls | Inspect native data and a channel observation; do not treat Catalog presence as placement. |
| A wrong source fact in any route | Product/variant record, policy or maintained compatibility source | Correct the owner-approved fact, then repeat the same question in the relevant route. |
These routes can coexist. This table is an investment choice, not a claim that one route reaches more shoppers or has a better return.
If you want to operate the conversation
A customer-facing assistant is something the merchant deploys and supports. It needs clear answers, a way to handle unknowns, and an owner for changes. Shopify says merchants can create a conversational agent with its Agent Platform and track questions routed through that agent in Shopify Knowledge Base. Knowledge Base also lets merchants review and customize store FAQs. Those are native places to start for a service or policy question. Shopify Knowledge Base.
Illustrative decision: if shoppers repeatedly ask “What is your return policy for opened equipment?”, review the published policy and FAQ, then decide whether a merchant-operated assistant would improve the on-site experience. StoreKnows is not that chatbot.
If shoppers bring an agent to your store
Shopify documents native WebMCP tools that let a compatible browser agent search products, get product and variant details, manage a cart and look up store policies. These tools are supplied on Liquid storefronts and Hydrogen developer preview; Shopify currently describes browser-agent support as limited to Chromium-based browsers. The agent chooses which available tool to call. Availability does not mean every shopper has such an agent or that it will use a particular fact. Shopify WebMCP.
Test the native response first. If it already returns the decisive product fact, another tool might add nothing. If an important maintained distinction still cannot be answered in a defined setup, a StoreKnows simulated check can help examine that question. A supported read-only StoreKnows tool may add selected merchant-approved facts for a compatible visiting agent. Compare actual responses before deciding to publish; neither registration nor a simulated improvement guarantees a consumer-agent answer. Read the native storefront tool guide for the mechanics.
If you want to appear in outside AI shopping
Shopify Catalog supplies eligible product data to connected channels and supports documented mapping of product-data sources. Shopify says the final display, rank and wording are controlled by the channel. A browser tool on your storefront is a different route; it does not rewrite a channel feed or guarantee a recommendation. Begin with product eligibility, clear maintained facts and the relevant channel controls. Shopify Catalog · Agentic Storefronts.
If the answer is a store FAQ, use Shopify Knowledge Base and the authoritative policy. If it is an exact product distinction, use the FAQ-versus-product-data diagnostic to find the source. The first purchase decision may be no purchase: correct native data and see whether the important question is already answered.
A one-question decision before spending
Write down one product or service question, why it matters, where the shopper asks it, the source your team maintains, and what the native route actually returned. Then ask: is the remaining gap in the merchant-owned conversation, in the product source, in outside distribution, or in the facts a visiting agent can call? A missing trace or unreliable reference means “investigate,” not “buy.”
StoreKnows is independently developed and not affiliated with, endorsed by or sponsored by Shopify or the AI platforms discussed.
Choose the route you need to test
Match the customer experience to the route before choosing a tool or changing product data.
Check the channel and source map