Agentic commerce on Shopify

Agentic commerce is shopping where AI agents act instead of just answering: they search, compare and sometimes buy for the shopper. For a Shopify store it has two sides. AI assistants such as ChatGPT can show your products to buyers and send them to your store to pay. Your own agent then sells to them in your store chat.

This guide covers both sides, with a source for every outside figure. It then uses our storefront agent, the Ambassador from Reponse, as a worked example. We make it, so test what we say against the criteria further down.

The two meanings of agentic commerce

Agents that shop for the buyer

Most definitions start on the shopper's side. Stripe describes AI agents that find, compare and potentially buy products for customers. Mastercard says the same: the agent completes the task with limited or no manual input from the user.

Where the buyer pays depends on the channel. The Shopify Help Center calls ChatGPT a "discovery-focused referrer platform": buyers find products there, then pay in your online store checkout. Microsoft Copilot and Google AI Mode and Gemini work differently for customers based in the United States. Those customers pay in a Shopify-powered checkout inside the chat. Elsewhere, Google sends buyers to your online store.

Shopify also notes that chat and loyalty blocks may not load in that in-chat checkout. A buyer who pays there doesn't see your store chat or your rewards.

The store's own agent

A second group of sources covers the merchant's side. IBM says agents act for consumers or for businesses, and lists merchant agents next to shopping assistants. McKinsey describes a shopper's agent dealing with a retailer's in-house AI commerce agent. In January 2026, Google launched Business Agent on Google Search: a branded virtual sales associate that eligible U.S. retailers switch on in Merchant Center.

For a merchant, this second meaning is agentic customer experience. An agent runs the store conversation on its own and moves the sale forward.

Where the label stops

Not every source agrees. Mastercard treats AI customer service agents as a separate field. In its agentic commerce guide, Shopify calls a chatbot that helps shoppers browse and decide conversational commerce. It keeps "agentic" for agents that do the shopping. In June 2025, Gartner warned about "agent washing": existing products, chatbots among them, rebranded as agents without substantial agentic capabilities.

Shopify's definition of an AI sales agent is broader. Where a basic chatbot follows scripted responses, a sales agent interprets context and takes a series of actions toward a sale. That is our working rule: an on-site agent earns the word when it acts. It searches the catalogue, checks price and availability with Shopify, asks which variant and builds the cart.

Why both sides matter to a Shopify merchant

The two sides meet in your store. Shopify says Agentic Storefronts is active by default for eligible stores. Eligibility includes a paid plan, completed store policies and products that ship to the US or Canada. Your products can then appear in AI channels such as ChatGPT, Copilot and Gemini.

So the buyer often arrives mid-conversation. In Q1 2026, Shopify's data showed that more than half of AI-referred sessions start on a product page, against about 20% for organic search. Adobe looked at Prime Day 2026 on U.S. retail sites. Visits from AI sources converted 40% better than non-AI channels such as paid search, email and social media.

Keep that in proportion. Adobe notes on the same page that AI traffic volume remains modest next to other major channels. Shopify says organic search still refers more sessions than all tracked AI platforms combined.

The point is the habit, not the volume. Shopify says these shoppers have already described their needs and narrowed their choice inside the AI platform. Our reading: the questions left are practical. Which size? Is it in stock? How does shipping work? Your store's agent can answer those in the chat, from your own catalogue and policies.

Agentic commerce vs traditional e-commerce

The usual comparison has two columns. On Shopify, a third one matters: the agent that works inside your store.

What AI shopping agents read

An AI assistant doesn't browse your store the way a person does. It reads structured product data, mostly through Shopify. Three sources show what that data contains.

The Universal Commerce Protocol (UCP)

In January 2026, Google announced the Universal Commerce Protocol, co-developed with Shopify, Etsy, Wayfair, Target and Walmart. In Google's words, UCP "establishes a common language for agents and systems to operate together across consumer surfaces, businesses and payment providers". The specification is open source. Shopify runs UCP servers that let an agent search products across Shopify merchants, build a cart and turn it into a checkout.

Our own agent uses the same protocol inside your store. Before it shows a product, it asks your store's catalogue for the live price and availability, and it builds the cart through UCP before handing it to your Shopify checkout.

What the catalogue tells an agent

Shopify's Global Catalog documentation lists what an agent receives for each product: the price, each variant's availability, a rating with its review count, and links to the seller's policies, including the refund policy. Some attributes, such as top features and unique selling points, are inferred by Shopify from your product data. Shopify says to treat them as "discovery and merchandising signals, not as merchant-authored source text".

OpenAI's product feed specification points the same way. It has fields for availability, the return policy and return window, and the review count and star rating. Raw reviews and customer questions are outside it.

Answers to the questions shoppers ask

Shopify's free Knowledge Base app generates facts about your store, such as shipping and returns, that AI shopping platforms can use when shoppers ask about it. You can review and edit them, add your own, and see which questions come from AI agents. Shopify presents it as a way to make answers about your store more accurate, not to make your store appear more often.

Your own help articles can play the same part, from your domain. Our help center, free in Reponse, can publish them as public pages of your store, at yourstore.com/apps/reponse/help, with their own sitemap and llms.txt for search engines and AI assistants. The store must not be password-protected, and each engine still decides what it indexes.

A checklist for your store

1. Complete your store policies. Refund, shipping, privacy, terms of service and contact information. Shopify counts completed policies among the conditions for Agentic Storefronts, and the catalogue links agents to them.
2. Give every product a clear title, images and a price, and keep variant stock up to date. Shopify asks for a title, an image and a price for Agentic Storefronts, and agents check availability variant by variant.
3. Write facts, not slogans. Materials, dimensions, sizing, what's in the box. Inferred attributes come from your product data, and an agent can only repeat what it finds there.
4. Answer shipping and returns once, clearly. In your policies, and in the Knowledge Base app if you sell through AI channels.
5. Check what AI can read on your store. Our free Shopify AI Readiness Checker reviews your catalogue, robots.txt, structured data and policies, and gives you a ready-made llms.txt.
6. Get your store chat ready for buyers who arrive with a question. Shoppers referred by AI often land on a product page with one practical question left. An agent in your store chat can answer it from your catalogue and policies, and hand over to your team when it can't.

What an AI sales agent does in a Shopify store chat

Here is what the Ambassador, the AI shopping assistant in Reponse's store chat, does in a Shopify store today.

It learns your store by itself

Once Reponse is installed, the agent learns your store by itself: your catalogue, pages and policies, read again regularly. There's nothing to write.

It searches before it says no

The agent searches your Shopify catalogue, synced automatically, before it tells a shopper that a product isn't available. It can search by name, type, brand or keyword.

It checks price and stock with Shopify

Before it shows a product card, the agent asks your Shopify store for the live price and availability. It uses the price of the shopper's own market, with a note on whether tax is included. It won't show a product card for anything Shopify reports as sold out.

It asks which variant

When a product comes in several sizes or colours, the agent asks which one the shopper wants. It shows a picker in the chat instead of guessing. It then checks the chosen variant's stock in your synced catalogue. If it's out of stock, the agent says so and suggests an alternative.

It builds the cart

Once the shopper has chosen, the agent builds the cart and applies the discount code of your running campaign, if there is one. It then hands a Shopify cart to your checkout.

Payment stays in Shopify

Payment always happens in your Shopify checkout. Reponse never takes payment itself. Your checkout, payments, theme, catalogue and orders stay in Shopify.

Rewards and referral, inside the chat

In the chat's Rewards tab, a signed-in customer sees their points, and their tier and referral link when you turn them on. If you let customers spend points, they can turn them into a single-use Shopify discount code tied to their account. One tap applies it to their cart, so the code carries through to your Shopify checkout. The Rewards tab needs no AI agent.

Already use another loyalty or referral app? Reponse recognises the main loyalty and referral apps on your storefront. It then pauses the matching part of its own programme, so shoppers don't see two programmes. You can change that choice in the settings.

It recognises returning customers

Customers signed in to your store are recognised through Shopify, so their chats are tied to the right customer. With an identified customer, the agent can keep what matters from earlier chats, such as sizes and preferences. Long-term memory is reserved for identified customers. An anonymous visitor who comes back on the same browser gets their current thread back, and nothing more.

It speaks English and French

The Ambassador talks to your shoppers in English and French, in the chat and in its replies.

Where humans stay in the loop

The AI runs a conversation on its own in one place: your store chat. There, the Ambassador answers shoppers 24/7. Around it, your team keeps control.

• Shoppers can ask for a person at any time. By default, one click opens a ticket for your team with the whole conversation. If the shopper isn't signed in, the chat asks for their email so your reply can reach them.
• Your team can take over. Anyone on your team can take over a chat from the Shopify admin. The AI pauses on that thread and picks up again when you hand it back.
• Support replies are drafts. In your admin, the Support Agent drafts replies and your team approves them.
• Spending has a ceiling you choose. After the trial, each AI task is billed $0.05 until the monthly budget you set is reached. If you set the budget to $0, or once it is spent, the AI hands over to your team until the next billing cycle.

Your team's work is never billed: Reponse live chat gives you unlimited human live chat, tickets and a shared inbox, with unlimited seats.

How to evaluate an agentic commerce solution for Shopify

Before you compare tools, separate the two layers. Agentic Storefronts is Shopify's distribution layer: it puts your catalogue in AI channels. An on-site agent handles the conversation inside your store. This section covers the second layer.

We make Reponse, so its column is our own description of how the product works today. Ask every vendor, us included, the same questions.

Then ask for evidence you can check. In June 2025, Gartner predicted that over 40% of agentic AI projects will be cancelled by the end of 2027. It cites escalating costs, unclear business value and inadequate risk controls. A conversion figure you cannot verify tells you little, whoever publishes it.

How to set up an AI sales agent on Shopify

With Reponse, setup takes four steps, all inside Shopify.

1. Install Reponse from the Shopify App Store. It opens inside your Shopify admin, and Shopify bills it: there's no card to enter.
2. Turn on the Reponse app embed in your theme editor. It takes one click and no code. The chat never appears on your cart, checkout or account pages.
3. Start the 14-day trial. Take the Reponse plan, at $0 a month, on Shopify's plan page. The trial covers up to 500 AI tasks; then each AI task costs $0.05, on your Shopify invoice.
4. Set the rules for your team. Choose who handovers go to by default. An owner or admin also sets the monthly budget for the AI tasks; at the end of the trial, Reponse proposes one, sized from the trial.

See the plan on the pricing page, or see the storefront agent on the Ambassador page.

FAQ

What is agentic commerce on Shopify?

Agentic commerce is shopping where AI agents take actions, not just answer questions. Most definitions describe assistants that search, compare and sometimes buy for consumers. On Shopify it has two sides. Agentic Storefronts lists eligible stores' products in AI channels such as ChatGPT, Copilot and Gemini. A store's own agent sells to shoppers in its store chat.

Is an AI chatbot on Shopify a form of agentic commerce?

Whether an AI chatbot on Shopify counts as agentic commerce depends on what it does. Shopify's agentic commerce guide calls a chatbot that helps shoppers browse and decide conversational commerce, and keeps "agentic" for agents that do the shopping. Shopify's definition of an AI sales agent is broader: unlike a scripted chatbot, it takes a series of actions toward a sale. An on-site agent that searches the catalogue, checks price and availability with Shopify, asks which variant and builds the cart meets that definition.

Can you add an AI chatbot to Shopify?

Yes. With Reponse, you install the app from the Shopify App Store. Then you turn on its app embed in the theme editor, in one click and with no code. Your team's live chat is never billed, and the AI agents come with a 14-day trial of up to 500 AI tasks, then $0.05 per AI task. Payment always happens in your Shopify checkout, and Reponse never takes payment itself.

Can shoppers still talk to a person?

With Reponse, shoppers can ask for a person at any time. By default, one click opens a ticket for your team with the whole conversation. Anyone on your team can also take over a chat from the Shopify admin, and the AI pauses on that thread.

Which languages does Reponse's storefront agent speak?

Reponse's storefront agent, the Ambassador, speaks English and French, both in the chat and in its replies.

How much does Reponse cost, and what counts as an AI task?

Reponse is one plan at $0 a month, billed on your Shopify invoice. A 14-day trial covers up to 500 AI tasks; a store that reaches them sooner sees the AI hand over to its team until the trial ends, or until it starts billing. Then each AI task costs $0.05, within the monthly budget you set; the current prices are on the pricing page. An AI task is one piece of work an agent completes: a conversation it leads, counted once however many messages follow, a ticket it sorts and drafts a reply for, a review it drafts an answer to, a help article it drafts. Your team's own work never counts.