What is Agentic Commerce?

Back to AgenticCommerce.co.nz

Why Agentic Commerce?

The motivation behind agentic commerce is clear:

• Seamless user experience: People already use conversational AI to get advice on what to buy. Agentic Commerce closes the loop, making it possible to go from intent → recommendation → purchase without leaving the chat.
• Capture high-intent demand: Instead of redirecting a user to an external site, merchants can meet buyers directly where the purchase intent happens.
• Standardization across platforms: Without a standard, every merchant would have to build custom integrations for each AI. ACP creates a single open protocol that works across agents.
• Maintaining merchant control and trust: Merchants remain the "merchant of record." They keep control over pricing, payment, fulfillment, and support, while the protocol simply provides a secure bridge.
• Security and authorization: ACP ensures AI agents act only with explicit, scoped user permission.

This approach reimagines commerce: AI agents don't just assist—they participate directly in the transaction lifecycle, safely and transparently.

What is the Agentic Commerce Protocol (ACP)?

The Agentic Commerce Protocol is the open standard that makes agent-driven transactions possible. Co-developed with Stripe, it lays out how merchants, AI agents, and payment providers exchange structured information to execute commerce in a trusted way.

Key Properties

• Open and interoperable: ACP is designed to work across multiple AI agents, merchants, and payment providers.
• Layered architecture: It includes specifications for product feeds, agentic checkout flows, and delegated payments.
• REST + JSON APIs: Merchants expose simple endpoints that allow agents to check inventory, initiate checkout, and finalize payment.
• Scoped, ephemeral tokens: Payments are authorized with temporary tokens that restrict amount, merchant, and timeframe. This prevents agents from going beyond user intent.
• Easy integration: For Stripe users, enabling ACP-powered payments may be as simple as a small code change. For others, Stripe can act as the intermediary payment layer.
• Minimal data exchange: Only the information required to fulfill the order is passed through the protocol.
• Flexibility: ACP is built to support not just one-off purchases, but subscriptions, asynchronous "buy later" flows, in‑store pickup, and dynamic pricing.
• Community-driven: The protocol is open source, licensed for broad use, and designed for community participation.

Walkthrough: The Agentic Commerce Flow

1. Product Discovery

1. A user asks an AI, for example, "Find me a wireless headset under $200."
2. The AI agent queries merchants' product feeds made available via ACP.
3. Products are ranked by relevance, not preferential treatment for merchants that support instant checkout.
4. The user sees eligible items with a "Buy" option.

2. Checkout

1. The user clicks "Buy."
2. The agent gathers shipping and payment details.
3. The agent sends a checkout request to the merchant's ACP endpoint.
4. The merchant validates inventory, shipping, tax, and risk.
5. The merchant returns a checkout session to the agent, possibly with options for the user to choose from.

3. Delegated Payment

1. The agent creates a delegated payment request, generating a token scoped to the transaction.
2. This token flows through the payment provider (often Stripe) for verification.
3. A verified token is returned to the agent and passed to the merchant to finalize payment.
4. The merchant captures funds and returns order confirmation.
5. The user sees a completed purchase confirmation in the agent interface.

4. Fulfillment & Lifecycle

Merchants then fulfill the order as they normally would—shipping, returns, refunds, and customer service all remain within their control. ACP doesn't replace existing commerce systems; it connects them to AI agents.

PXM and Agentic Commerce

Product Experience Management (PXM) plays a crucial role in enabling effective agentic commerce. As AI agents become more sophisticated in understanding and presenting products to users, the quality of product data, imagery, and contextual information becomes paramount. What is PXM? PXM solutions help merchants maintain rich, accurate product information that AI agents can leverage to provide better recommendations and seamless purchasing experiences. This synergy between PXM and agentic commerce creates a more intelligent, context-aware shopping environment where both human and AI interactions benefit from comprehensive product intelligence.

Partnership Perspective: OpenAI and Stripe

OpenAI and Stripe have partnered to launch ACP and are rolling it out through Instant Checkout in ChatGPT. Stripe has emphasized that ACP is built to be business‑friendly: merchants integrate once, keep full control over their brand, pricing, and risk models, and immediately become compatible with multiple AI agents.

Stripe has also noted that:

• Ease of adoption is key: Stripe merchants may enable ACP-powered payments with minimal effort.
• Flexibility matters: The protocol supports not only simple digital purchases but also complex transactions like subscriptions, physical goods, in‑store pickups, and asynchronous "buy later" purchases.
• Scalability is essential: A single ACP integration can work across all participating AI agents, reducing fragmentation.
• Trust and fraud protection are central: Merchants need to reliably distinguish authorized agent transactions from bad bots, and ACP's design reflects that.
• Real deployments are underway: ACP is already powering ChatGPT's Instant Checkout, with early support for U.S. Etsy sellers and plans to expand to millions of Shopify merchants.

This positions ACP not as a theoretical protocol, but as an actively deployed system shaping real-world commerce.

What ACP Enables

For Users

• Frictionless purchases directly in chat.
• Clear consent at every step.
• A single interface for discovery and buying.

For Merchants

• Access to a new channel where customers are already making buying decisions.
• Full control over transactions, pricing, fulfillment, and customer relationships.
• Low integration overhead, particularly for Stripe users.
• The ability to reach multiple AI agents with one integration.

For AI Agents

• Native commerce capabilities without needing to act as a merchant.
• Compatibility across platforms through a shared standard.
• A focus on orchestration while leaving fulfillment and risk to merchants.

Challenges & Future Directions

There are still challenges ahead:

• Security and trust: Ensuring only authorized purchases go through.
• Liability: Clarifying responsibility in the event of fraud or mistakes.
• Adoption: Merchants and platforms need to embrace ACP for it to scale.
• Competition: Other protocols, such as those proposed by Google, may emerge.
• Complexity: Handling multi‑merchant carts, subscriptions, or international compliance adds difficulty.
• User confidence: Building trust so users are comfortable letting agents complete purchases.

Summary

Agentic Commerce envisions AI agents that don't just advise but also transact on behalf of users. The Agentic Commerce Protocol (ACP), co‑developed by OpenAI and Stripe, makes this possible by defining open, secure standards for product feeds, checkout, and delegated payments.

Users get a seamless, trusted buying experience; merchants gain a powerful new channel without losing control; and AI agents expand their usefulness by connecting intent to action. With ACP already in production and expanding, it is positioned as one of the first real frameworks for a future where commerce is increasingly agent‑driven.