Join the Community

24,213
Expert opinions
40,775
Total members
355
New members (last 30 days)
216
New opinions (last 30 days)
29,305
Total comments

Agentic Payment Protocol solves important problems that enable agentic commerce

Large language models have simplified how people use the internet. People do not have to sift through mountains of internet links anymore. They just have to ask the chat interface, which answers based on the content it has crawled from those pages in the past. This means the webpages do not serve the users directly anymore. This breaks how websites manage to detect users and personalize their experience. In short, the identity aspect of the user has gone away and instead it has moved to the LLM.
 
However, LLMs are often not querying any websites live. They do know the user's identity, but they are often answering based on training data and model weights. Even in the case of features like deep research or grounding, the LLM might be using data obtained from other websites, but it does not have any ability to provide user identity to those websites.
 
This is acceptable for static content, but when it comes to agentic shopping, it might be beneficial if there was some way the agent could authenticate the user with the shopping websites. This problem gets tricky as the agent's strength relies on reaching out to a large number of shopping websites; however, the user might want to be picky about their identity information being delegated to other websites.
 
Further, as the agent decides to buy stuff automatically, the user identity and possibly payment authorization information might have to be passed to the merchants as well. The merchants need to be certain that this is indeed the user's intention and not agentic hallucination.
 
The Agentic Payments Protocol (APP) aims to address this important concern.
 
It provides three important features as its primary core features:
 
• Authorization: Proving that a user gave an agent the specific authority to make a particular purchase.
• Authenticity: Enabling a merchant to be sure that an agent's request accurately reflects the user's true intent.
• Accountability: Determining accountability if a fraudulent or incorrect transaction occurs.
• Trust building
 
AP2 solves this problem through the notion of "mandates". A mandate is a cryptographically signed instruction by the user. These are digitally signed and are tamper-proof. Further, these mandates can be linked with payment credentials. Together, there is a clear, auditable trail of what the user agreed to pay and what method was used to pay at a certain point in time.
 
In the context of payments, payments are often of two types. In-session payments, where the user is expected to be in session and make the final call, or delegated payments (such as subscriptions) where the merchant can charge the user through a card on file or some form of mechanism that does not require the user to be in session. AP2 can deal with both these scenarios.
 
Agentic Commerce and Agentic Payments
 
The AP2 protocol provides a solid foundation to enable a user's shopping agent to communicate with a merchant agent without the need for even an e-commerce shop. This pure agent-to-agent communication can handle the entire process of shopping with only basic intent from the human consumer.
 
If agentic commerce becomes mainstream, it has the potential to transform how people shop and how sellers sell. Sellers can focus on what they are good at, such as making stuff. They will have to worry less about creating a website, handling payments, running ads, and so on, as long as they have a good selling agent that can handle managing the sale.
 
Time will tell if agentic commerce will take off or not, but AP2 gives companies a solid foundation to build the basic building blocks.

External

This content is provided by an external author without editing by Finextra. It expresses the views and opinions of the author.

Join the Community

24,213
Expert opinions
40,775
Total members
355
New members (last 30 days)
216
New opinions (last 30 days)
29,305
Total comments

Now Hiring