What Is Agentic Commerce? How to Sell Your Products in ChatGPT

Last updated: November 2025

Key takeaways:

• Agentic commerce lets AI agents like ChatGPT find products, compare options, and complete checkout for shoppers without ever leaving the chat.
• Two open standards make it work: the Agentic Commerce Protocol (ACP) covers checkout and payment flows, while the Model Context Protocol (MCP) connects agents to tools and data.
• Merchants can prepare now by cleaning up their product feed, applying for ChatGPT checkout eligibility, and integrating a compatible payment processor.
• High-risk merchants face additional underwriting, fraud, and chargeback considerations before agent-initiated sales can flow smoothly.

The world of commerce is changing fast, and artificial intelligence is leading the charge. One of the most exciting developments is something called agentic commerce — a new way for consumers, merchants, and businesses to interact through smart agents that handle everything from shopping and checkout to payment processing and customer service.

At its core, agentic commerce combines AI, open standards, and natural language processing to make the shopping experience smoother and more personal. Imagine talking to ChatGPT in an online chat window, describing what you want to buy, and having a personal shopper agent instantly find the right product, compare prices, and handle your payment — all in real time.

What Is Agentic Commerce?

Agentic commerce is a model of online shopping in which an AI agent — such as ChatGPT — finds products, compares options, and completes checkout on a customer's behalf. Instead of clicking through a website, the shopper describes what they want in a conversation, and the agent handles discovery, payment, and order confirmation inside the chat.

This is possible thanks to a set of shared rules and technologies, most notably the Agentic Commerce Protocol (ACP) and the Model Context Protocol (MCP). These protocols enable AI agents to connect across different infrastructures, APIs, and payment processors. That interoperability allows merchants to reach more customers and participate in a broader ecosystem of digital commerce — no matter what technology or payment system they use.

For merchants, agentic commerce represents the next major leap in programmatic commerce, where automated agents can initiate, verify, and complete agentic payments on behalf of users, all within an LLM.

How Agentic Commerce Works in ChatGPT

ChatGPT is a prime opportunity to leverage agentic experiences. When consumers chat with the model, the digital agent can access inventory, suggest personalized recommendations, and even handle checkout securely. These shopping agents leverage open APIs and AI-driven personalization to make every customer interaction feel effortless.

For instance, if a shopper asks about skincare products, the AI agent could pull in product information, compare brands, and complete a payment through an integrated payment processor — all inside the same chat. The entire process happens through secure credentials, encryption, and open source systems that help ensure trust and data protection.

The Agentic Commerce Protocol (ACP) vs. the Model Context Protocol (MCP)

The two protocols behind agentic commerce sound similar, but they solve different problems — and understanding the difference helps you know what to build first.

The Agentic Commerce Protocol (ACP) is an open specification published by OpenAI that standardizes the commerce side of the conversation: how an agent surfaces a merchant's products, shares order details, and hands off payment information at checkout. Under this model, the agent facilitates the purchase, but the order is passed back to the merchant, who processes the payment through their existing systems and fulfills the sale.

The Model Context Protocol (MCP) is an open standard, originally introduced by Anthropic, that standardizes how AI models connect to external tools, data sources, and services. In a shopping context, MCP-style connections are what allow an agent to check live inventory, look up an order status, or call a merchant's API in the first place.

In short: MCP handles the connections and context an agent needs, while ACP handles the checkout and payment flow itself. Because both are open standards, a merchant who integrates once is positioned to participate across multiple agent platforms rather than being locked into a single one.

How Merchants Can Get Their Products Into ChatGPT

Understanding the concept only matters if you can act on it. Exact requirements will evolve as the ecosystem matures, but the path for merchants generally looks like this:

Step 1: Prepare Your Product Feed

Agents can only recommend what they can read. Build a clean, structured product feed with accurate titles, descriptions, pricing, availability, and images. The better your product data, the easier it is for an AI agent to match your catalog to what a shopper asks for.

Step 2: Apply for ChatGPT Checkout Eligibility

OpenAI outlines how merchants can express interest in agentic checkout in its developer documentation. Expect an application and review process: eligibility criteria, supported product categories, and availability are controlled by the platform and may change as the program expands, so review the current requirements before you build.

Step 3: Integrate a Compatible Payment Processor

Agent-initiated orders still settle through your own payment infrastructure, so you need a processor that can support the volume and risk profile of a new sales channel. That includes card processing and, depending on your business model, options like ACH payment processing. If your industry is considered high risk, this step deserves extra attention — more on that below.

Step 4: Test the Buy Flow

Before you count on agent-driven revenue, test the experience end to end: product discovery, checkout, payment confirmation, receipts, and post-purchase support. Confirm that orders flow into your fulfillment systems correctly and that your billing descriptor is recognizable, since customers who do not recognize a charge are more likely to dispute it.

Agentic Commerce for High-Risk Merchants

Here is the angle most coverage of agentic commerce skips: what it means for merchants in high-risk verticals.

Consumer-facing high-risk categories — think supplements and nutraceuticals, subscription products, and other card-not-present businesses — sell exactly the kinds of products shoppers already ask AI agents about. Whether a given category can participate in agent-initiated checkout, however, depends on each platform's eligibility rules and the card networks' requirements, so check the current guidelines for your vertical before investing in an integration.

Underwriting and merchant category coding also matter more when an AI agent initiates the payment. Orders arriving from a brand-new channel can look unusual to risk systems, and a sudden surge of agent-driven transactions can trip fraud and velocity rules if your account was not set up for that channel or volume. Properly established high risk merchant accounts — with accurate category codes, realistic processing volumes, and risk controls sized to your business — help keep agent-driven sales from being flagged or declined.

Payments, Fraud, and Chargebacks in Agent-Initiated Checkout

The most common merchant objection to agentic commerce is a fair one: when an AI agent completes a purchase, the transaction is card-not-present by definition, which raises questions about liability, disputes, and fraud screening.

Agentic Payments and Fraud Screening

Agentic payments are designed around secure credentials, encryption, and open standards rather than stored passwords or screen-scraping. Even so, merchants should treat agent-initiated orders like any other card-not-present sale: keep fraud screening in place, monitor velocity as the new channel ramps up, and watch early transactions closely so legitimate agent-driven orders are not mistaken for suspicious activity.

Chargebacks and Disputes in Agentic Commerce

In the ACP model, the order is passed to the merchant, so a purchase completed by an agent still settles through your own processing relationship — and disputes follow the familiar card-network chargeback process. What changes is your evidence: order confirmations, clear product data, and a recognizable billing descriptor matter even more when the customer never visited your website. Building chargeback prevention into your setup before you turn on agent-driven sales helps protect both your revenue and your processing relationship.

Why Businesses and Merchants Should Care About Agentic Commerce in LLMs

For businesses and merchants, the benefits go far beyond convenience. Agentic commerce opens new opportunities for marketing, brand strategy, and revenue growth. AI agents can analyze consumer behavior, optimize pricing, and personalize content in real time. Sales that would otherwise be interrupted by customers needing to navigate out of ChatGPT to your company's site can happen seamlessly within the chat, increasing the potential for sales of your goods and services.

This technology also improves customer service by allowing agents to answer questions, process returns, and provide post-purchase support within the same chat window. With this kind of automation and personalization, businesses can deliver better experiences while reducing operational tasks and improving overall efficiency.

The Next Chapter of Programmatic Commerce

The future of commerce will be driven by AI agents and intelligent payment systems that make shopping faster, smarter, and more personalized. Through agentic commerce, merchants can finally combine the best of technology, marketing, and customer experience — turning every online chat into an opportunity for growth and sales.

As the agentic commerce landscape continues to expand, PayKings wants to hear from you. Contact us today to let us know if you'd like PayKings to add agentic commerce in ChatGPT as a service we can provide to you.