AI use case

Agentic checkout for merchants, selling through third party AI shopping agents

Agentic checkout for merchants is how a retailer lets third party AI shopping agents read its catalog, build a cart and take payment while it stays merchant of record. It covers the product data, the cart and checkout endpoints and the order handling behind them.

By Len Debets · Last verified 14 October 2026 · 6 public deployments

100%
Reported conversion uplift
Shopify, organization claim.
USD 7500 to USD 140,000
Indicative value per year
A specialty online retailer with USD 50 million in annual online sales. Worked example, see how it is calculated.

What problem does it solve?

Shoppers have started their searches in AI assistants. Adobe Analytics measured traffic to U.S. retail sites from generative AI tools up 693.4% in the 2025 holiday season against the year before, which Adobe describes as a modest base of users. An assistant can describe a merchant's products from scraped pages or third party feeds, and Shopify says searches using its own catalog data converted at twice the rate of searches using scraped data. Without a connection, the sale happens, if at all, after a click back to the merchant's site.

Opening the store to the agents takes product data an agent can trust and a cart and checkout an agent can call. A purchase made somewhere else still needs the usual order, tax, shipping and returns handling. The merchant also decides where the purchase happens. Inside the assistant, the shopper never visits the merchant's site. Through a merchant app inside the assistant or a click out, the merchant keeps its own checkout and customer relationship.

The first year produced a mixed record. Etsy was the first live partner for ChatGPT's Instant Checkout in September 2025. In March 2026 an OpenAI spokesperson told Digital Commerce 360 that "Instant Checkout is moving to Apps, where purchases can happen more seamlessly." Walmart's Daniel Danker, who oversees design and product, told WIRED that conversion for items sold inside the chatbot ran three times lower than for items that sent the shopper to Walmart's site.

Instacart built a full shopping app in ChatGPT, and Shopify reports that AI driven orders tripled year over year while remaining small against its total volume. This page covers the merchant's side of that arrangement. The agent that pays for the shopper is covered by the agentic payment initiation page, and a merchant's own assistant by the conversational shopping assistant page.

How does it work?

  1. Publish product data an agent can use. The merchant exposes its catalog through a feed or API with live price, stock, variants, shipping options and return terms. Shopify calls its version Catalog and reports that AI searches powered by it converted at twice the rate of searches using scraped data. Instacart's launch release lists up to date inventory and pricing among what its ChatGPT app offers.
  2. Connect to the protocols and surfaces that carry traffic. Two open specifications are in use: the Agentic Commerce Protocol, maintained by OpenAI and Stripe, and the Universal Commerce Protocol, which Google says it developed in collaboration with industry leaders including Shopify, Etsy, Wayfair, Target and Walmart (ucp.dev lists more partners). Microsoft runs Copilot Checkout with PayPal, Shopify and Stripe. A merchant on a hosted platform usually switches these on through the platform.
  3. Expose cart and checkout. The agent creates a cart, the merchant prices it (tax, shipping, discounts, availability) and returns the total, and the shopper confirms. Payment goes through a delegated credential or token with the payment provider, and the merchant stays merchant of record with its own business rules.
  4. Choose where the purchase completes. Options seen so far: checkout inside the assistant, a merchant app inside the assistant (Instacart's app in ChatGPT, Etsy's app in ChatGPT), or a click out to the merchant's own checkout.
  5. Handle the order like any other. The order lands in the order management system with a channel tag, goes through fraud screening, fulfillment, customer service and returns, and is attributed to the agent that sent it. Shopify added an agentic section to its merchant admin in May 2026 for this cross channel attribution.
  6. Measure and adjust. Compare agent referred traffic, conversion, order value and returns with other channels, and move the checkout location or the surface mix accordingly.
Audience
Customer facing
Autonomy
Supervised agent
Adoption
Emerging
Channels
Web chat, Mobile app, API and system to system

What is it worth?

Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.

Value benchmarks for Agentic checkout for merchants, selling through third party AI shopping agents
KPIMedianReported rangeData pointsClaimed by
Conversion upliftToo few to pool
-67% to 100%
22 organization

Value drivers: Revenue growth, Customer experience.

Indicative value

A specialty online retailer with USD 50 million in annual online sales

USD 7500 to USD 140,000

Incremental gross profit from agent originated sales per year

How this is calculated

Formula: onlineSales * agentShare * incrementalShare * grossMargin. The low scenario uses every low input, the high scenario every high input.

InputLowHighBasis
Annual online sales onlineSales, USD per year50,000,00050,000,000The reference retailer.
Share of online sales that originates with AI shopping agents agentShare, fraction of sales0.0020.01Editorial assumption. Etsy's CEO put agentic traffic below 1% of total traffic in the fourth quarter of 2025, so the high value is set at that ceiling, and Shopify says AI volume is still small against its total GMV. Replace with your own referral data. Source
Share of agent originated sales that would not have happened otherwise incrementalShare, fraction of agent sales0.30.7Editorial assumption. Part of this volume would have reached the merchant through search or direct visits. Replace with a holdout or attribution test.
Gross margin on those sales, after payment and platform fees grossMargin, fraction of sales0.250.4Editorial assumption. Replace with your own margin after fees. WIRED reported that OpenAI bet it could earn revenue by charging a commission on purchases made through ChatGPT, whereas Shopify says agentic transactions carry the same economics as an online store transaction.

What it leaves out: Gross profit only. It leaves out the cost of integration and of running the feeds, fraud and chargeback losses on agent orders, the return rate of these orders, and the risk that agent traffic grows faster than the editorial range above.

Who already uses it?

6 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.

Shopify

Canada · Retail and ecommerce · 2026

ProductionGrade B

Shopify gives its merchants an agent facing layer: a Catalog of product data for AI discovery, the Universal Commerce Protocol it helped develop so agents can access product data, create carts and complete checkout within each merchant's rules, and an agentic section in the merchant admin with cross channel attribution (launched May 2026). On its second quarter 2026 call Shopify said AI driven traffic and orders to Shopify stores tripled year over year, from a base that is still small against total GMV, and that AI searches powered by Catalog converted at twice the rate of searches using scraped data.

  • Conversion uplift: 100%, second quarter 2026
    "We're seeing that AI searches powered by Catalog converted twice the rate of those using scraped data."
    Claimed by: organization

Etsy

United States · Retail and ecommerce · 2025

ProductionGrade B

Etsy was the first live partner for ChatGPT's Instant Checkout in September 2025 and later expanded integrations with Microsoft Copilot (where its sellers' items are available in Copilot Checkout) and Google, plus an agentic payments agreement with Stripe.

Etsy built an Etsy App in ChatGPT, which its first quarter 2026 shareholder letter describes as aligned with OpenAI's shift for agentic shopping to focus on retailer run apps (an OpenAI spokesperson said in March 2026 that Instant Checkout was moving to Apps). On its fourth quarter 2025 call the CEO said agentic traffic grew roughly 15 times year over year but still accounts for less than 1% of total traffic.

PYMNTS, reporting the call, wrote that these users showed higher intent and higher average order values and that users typically flow from AI recommendations to Etsy listings before completing a purchase.

No outcome disclosed.

Instacart

United States · Retail and ecommerce · 2025

ProductionGrade B

On 8 December 2025 Instacart launched an app inside ChatGPT that builds a grocery cart from a conversation, lets the shopper review it and pays through Instant Checkout, built on the Agentic Commerce Protocol with Stripe processing the payment. The release says the items are then shopped and delivered through Instacart's platform, and describes a partnership with more than 1,800 retail banners. In 2026 it extended the same approach to other AI platforms: management said on the first quarter call that it had integrated with ChatGPT and most recently Claude, and its second quarter release says it became Google's first grocery partner to integrate with Gemini. No order volume or conversion figures from these channels are disclosed.

No outcome disclosed.

PayPal

United States · Payments and cards · 2025

ProductionGrade B

In November 2025 PayPal enabled U.S. users of Perplexity's shopping experience to check out with PayPal without leaving the answer engine, from merchants such as Abercrombie & Fitch, Ashley Furniture, Fabletics, Adorama and Newegg. PayPal's agentic commerce services sync merchant catalogs and let merchants accept agent originated payments with their existing PayPal setup, while the retailer stays merchant of record and transactions pass PayPal's fraud screening, identity verification and purchase protection. No volumes or conversion figures are disclosed.

No outcome disclosed.

Walmart

United States · Retail and ecommerce · 2025

Paused or rolled backGrade B

Since November 2025, Walmart made about 200,000 products available directly inside ChatGPT through OpenAI's Instant Checkout, letting shoppers give OpenAI their shipping and payment details and place an order without leaving the chat. Daniel Danker, Walmart's EVP for design and product, told WIRED in March 2026 that conversion for purchases completed inside the chat ran at roughly one third of the rate for the same products when ChatGPT instead routed the shopper to walmart.com to check out, and he called that standalone experience inside the chat "unsatisfying". Walmart is instead embedding its own Sparky assistant inside ChatGPT, where shoppers log into Walmart and order with the payment and delivery details already saved there.

  • Conversion uplift: about -67%, November 2025 to March 2026, products sold directly inside ChatGPT via Instant Checkout
    "Conversion rates—the percentage of users following through with a purchase of an item shown to them by ChatGPT—have been three times lower for the selection sold directly inside the chatbot than those that require clicking out, according to Daniel Danker, who oversees design and product for Walmart."
    Claimed by: organization

Ulta Beauty

United States · Retail and ecommerce · 2026

AnnouncedGrade C

Google Cloud lists Ulta Beauty as rolling out agentic commerce inside AI Mode in Google Search and the Gemini app, where shoppers get Ulta Beauty product recommendations, compare options and complete checkout for eligible purchases within Google's conversational interfaces. The entry was added in the April 2026 edition of Google's list and describes a rollout, not results. No outcome figures are disclosed.

No outcome disclosed.

How do you implement it?

A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.

Data you need

  • Product catalog with accurate titles, attributes, variants and images, and live price and stock
  • Shipping rules, delivery promises and return policy as structured data
  • Tax calculation for the destinations the agents may ship to
  • A way to tag an order with the agent or surface it came from

Systems to integrate

  • Commerce platform or order management system
  • Payment provider with support for delegated payment tokens
  • Product information management or feed generation
  • Fraud screening and customer service tools that can see the order channel

Complexity: Medium

The build needs a catalog feed or API with live price and stock, cart and checkout endpoints that can price an order on request from an agent, and a payment provider that supports delegated tokens. Merchants on Shopify get much of this from the platform, which says it takes the agentic strategy off the merchant's plate. Custom stacks build the endpoints, and any merchant needs the order, tax and returns handling to treat agent orders like its own.

  1. 1

    Audit the product data before anything else

    Check price and stock accuracy, variant completeness and image quality for the products you want agents to sell. An agent reproduces what the feed says.

  2. 2

    Pick one surface and one checkout model

    Start with the assistant that already sends you traffic and test one model, either a merchant app, an in assistant checkout or a click out. Compare conversion against your own site.

  3. 3

    Test the full order path with real orders

    Run purchases from cart to delivery, return and refund on a small product set. Check tax, shipping, split shipments and the customer's receipt.

  4. 4

    Tag and attribute agent orders

    Record which agent and surface each order came from, so conversion, order value, returns and fraud can be read per channel.

  5. 5

    Set limits and widen in steps

    Cap order value and quantity for agent orders at the start, and raise the limits as returns and chargebacks stay in line with other channels.

Guardrails

  • The price and stock an agent shows must come from the same source as the merchant's own site
  • The shopper confirms the cart and the total before payment is taken
  • Order value, quantity and shipping destination limits for agent originated orders
  • Fraud screening on agent orders that can see the agent identity and the delegated credential
  • Customer service and returns work the same way for agent orders as for site orders

KPIs to instrument

  • Agent referred traffic, conversion and order value, compared with search and direct channels
  • Share of agent orders completed inside the assistant versus on the merchant's site
  • Return rate, chargeback rate and fraud rate on agent orders
  • Price and stock mismatches between the agent's view and the live site
  • Customer contacts per agent order

Human in the loop

A commerce lead owns which surfaces and products are open to agents and reviews agent order quality weekly: conversion, returns, chargebacks and customer contacts. Customer service handles disputed or mismatched orders with the agent channel visible on the order. Changes to the limits or to the checkout model need that owner's sign off.

Common failure modes

Conversion falls when the purchase moves into the assistant
Walmart reported items sold inside ChatGPT converting at a third of the rate of click outs. Test the checkout model against your own site before committing volume to it.
The surface changes under the integration
An OpenAI spokesperson said in March 2026 that Instant Checkout was moving to apps, about six months after Etsy went live as its first partner. Etsy then built an Etsy App in ChatGPT, which it describes as aligned with that shift. Build on open protocols and keep catalog and order handling independent of any one assistant.
Stale or wrong product data in agent answers
An agent quotes a price or availability that the checkout then contradicts. Serve the feed from the live source and monitor mismatches.
Agent orders reach support with no context
The service team sees an order with no sign of how it was placed. Pass the channel and the agent conversation reference into the order record.

What are the risks and rules?

EU AI Act

Depends on design

Feeds and endpoints alone are minimal risk. A merchant that provides its own conversational agent (its own LLM app or assistant, inside an AI platform or on its channels) takes the Article 50(1) transparency duties for that system: people must be told they are interacting with an AI system, unless that is obvious from the circumstances. Where the shopper talks to a third party assistant such as ChatGPT, the assistant is run by its provider, who carries its own obligations.

Rules that apply

Guidance

  • Agentic Commerce Protocol specification (OpenAI and Stripe, Global). Open standard for connecting buyers, their AI agents and businesses to complete purchases, maintained by OpenAI and Stripe.
  • Universal Commerce Protocol (Google and partners, Global). Open standard for discovery through checkout across agents and merchants, built so that businesses remain the merchant of record, with payment mandates for consent.
  • Consumer rights directive (European Commission, Europe). Sets the EU right of withdrawal and information duties for distance purchases, which apply to an order placed through an agent as to any other online order.

Controls to put in place

  • Order record that stores the agent, the surface, the shopper's confirmation and the delegated credential reference
  • Payment handled only through the payment provider's tokens, so card data stays with the payment provider and out of the merchant's PCI scope
  • Monitoring of price and stock mismatches between feeds, agents and the live site
  • Return and refund rules applied identically to agent orders
  • Disclosure of AI use in any merchant run shopping agent, unless it is obvious to the shopper

Frequently asked questions

Should a merchant complete the purchase inside the assistant or on its own site?

The evidence points to keeping the merchant's checkout in play. Walmart reported conversion three times lower inside the chatbot than for click outs, an OpenAI spokesperson said in March 2026 that Instant Checkout was moving to apps, and PYMNTS reported that Etsy sees shoppers typically move from an AI recommendation to an Etsy listing before completing the purchase. Instacart's app inside ChatGPT is a merchant run app with checkout, so both can coexist.

How much sales volume comes through AI agents?

Little so far, and growing fast. Etsy's CEO said agentic traffic was below 1% of total traffic while growing roughly 15 times year over year, and Shopify said the volume is small against its total GMV while AI driven traffic and orders tripled. Adobe measured a 693.4% rise in traffic to U.S. retail sites from generative AI tools in the 2025 holiday season.

Which protocols should a merchant support?

Two open specifications carry most announcements: the Agentic Commerce Protocol from OpenAI and Stripe and the Universal Commerce Protocol from Google with Shopify, Etsy, Wayfair, Target and Walmart. A merchant on a platform such as Shopify gets support through the platform; others should check which assistants their payment provider and commerce platform already connect to.

Do agent sales cost more to take?

It depends on the surface. Shopify's management said agentic transactions carry the same economics as an online store transaction. WIRED reported that OpenAI had made a bet that it could boost revenue by charging a commission on purchases made through ChatGPT.

How to cite this page

Blits.ai AI Use Case Library, "Agentic checkout for merchants, selling through third party AI shopping agents", last verified 14 October 2026, https://www.blits.ai/ai-use-cases/agentic-checkout-for-merchants. Licensed under CC BY 4.0. Method: how we verify use cases.

Changelog
  • 14 October 2026: First published

Related use cases

Payments and cardsBanking

AI agent for payment initiation within a customer mandate

Agentic payment initiation is an AI agent that initiates and completes payments or purchases on a customer's behalf, within a mandate the customer set in advance. The mandate sets spending caps, allowed merchants or categories, a tokenized credential and rules for when to ask for confirmation, and the agent then confirms and reconciles every transaction it made.

Deployments
17 public, best grade B
Reported conversion uplift
about -67%
Walmart, organization claim
Cross industryRetail and ecommerce

AI shopping assistant for product discovery and recommendations

An AI conversational shopping assistant answers product questions and recommends items from a retailer's own catalog, for a need, project or occasion a shopper describes. It works on the retailer's site or app, compares items grounded in the retailer's product data, reviews and stock, and hands the shopper to a basket, a store or a human expert.

Deployments
12 public, best grade B
Median conversion uplift
100%
3 deployments
Cross industryRetail and ecommerce

AI agent for order status, delivery changes and returns

An AI order status and returns agent answers "where is my order", changes delivery details and arranges returns, exchanges and refunds for online and omnichannel shoppers. It does this end to end, by reading and writing to the order, carrier and returns systems within the retailer's policy, and hands exceptions such as damaged goods, disputes and upset customers to a person.

Deployments
9 public, best grade B
Median customer satisfaction, vendor reported
92.5%
4 deployments
Retail and ecommerce

AI order fraud screening for ecommerce checkout

AI order fraud screening decides, at the moment an online order is placed, whether to accept it, hold it for review or decline it, for an online retailer or marketplace. It draws on the buyer's device, behavior, identity and order history, so genuine orders are approved quickly and fraudulent ones are stopped before they ship, often backed by a vendor guarantee against the chargebacks that slip through.

Deployments
9 public, best grade C
Reported revenue uplift
6%
Mango, vendor claim
Retail and ecommerceCross industry

AI product content and catalog enrichment for online retail

AI catalog enrichment is AI that drafts and repairs product content, such as titles, descriptions and alt text, for online retailers at catalog scale. It also extracts missing attributes such as color, size and material from supplier text and product images, and checks its own output before the content is published to the store and to search engines. A human owns the rules, the quality thresholds and the exceptions.

Deployments
6 public, best grade B
Reported accuracy
about 75%
Zalando, organization claim