Agentic Commerce
While the jury is still out on an industry definition of “agentic commerce,” we think of agentic commerce as the buying and selling of goods and services involving a non-human agent which helps facilitate the payment transaction.
Similar to the emergence of ecommerce and the subscription economy, agentic commerce introduces its own opportunities and challenges to the payment ecosystem. Opportunities include additional convenience for consumers and potential new exposure for merchants. Challenges include increased fraudulent activity and lost customer visibility.
Agentic payments have not seen the traction we all thought it might over the last year; however, the frameworks are being built in the background. Like other alternative payment methods introduced in the United States, agentic payments is leaning towards card issuance, and in turn, additional BINs for merchants to sort through.

“Merchants need a seat at the table when the current framework is adapted for agentic commerce to ensure it works for everyone and includes AI platforms as potential liability holders - there’s a lot of work to be done here and a lot more for merchants to lose with inaction.”
- Robbie MacDiarmid, CMSPI
Data as a Solution
The introduction of classifications like Customer-Initiated Transactions (CIT) and Merchant-Initiated Transactions (MIT) to the conversation has brought additional expertise and optimization to merchants’ payment strategies across the globe. Why? Because how payments are initially made has real implications on authorization performance and cost impacts.
In a similar fashion, when an agent facilitates a payment on behalf of a consumer, there are implications regarding liability, additional fees, and intent. All parties involved in the transaction need to be on the same page regarding the agent's involvement in order to clearly communicate the terms and potential liability of the payment. We suggest an additional classification, namely Agent-Initiated Transaction (AIT).
Utilizing the POS Environment part of the POS Data Code data element, a new value could be added to signal when a transaction is being initiated through an agent’s platform.
DE22: POS Data Code Character Examples
Below are examples of the four characteristics included in the POS Data Code data element. Card Reading Method can be used to distinguish between a Customer Initiated Transaction (CIT) and a Merchant Initiated Transaction (MIT). Revisions to this field could allow more data to be submitted regarding transactions involving AI agents making payments on behalf of a consumer. Including an Agent-Initiated Transaction (AIT) flag would increase visibility for merchants and improve the customer experience for both humans and their agents.
A Global Perspective
Click the regions below to read more.

APAC

Europe
What Should Merchants Do Now?
Merchants must demand visibility into agent-initiated transactions. Without clear identification of when an AI agent participates in a payment, managing liability, performance and customer outcomes will become increasingly difficult.
ISO 8583 Connection
DE 22: POS Data Code
The twenty-second data element (DE22) of ISO 8583 is the POS Data Code. The POS Data Code provides information surrounding how the transaction was completed. It is made of four parts including the card reading method used, the cardholder verification method used, the POS environment, and security characteristics.
Get Involved
Change only happens when people act. CMSPI and the MPL Council are working together to provide practical solutions to move the payments industry forward. Join one of our working groups to be a part of the conversation.
Working Group Interest Form
Loading...