Die MCP-Funktion von WeChat Pay ist online gegangen. Die Zahlungsproduktmanager werden wieder in die Aktion geraten. Zwei Richtungen, in denen die PMs tätig werden können.

Recently, the MCP of WeChat Pay has been updated. Compared with the MCP of Alipay, WeChat Pay is still in the experimental version. For example, currently it only supports user payments and does not support refunds.

Actually, the best application scenario for Alipay's MCP is AR glasses.

I once saw a video mentioning that Alipay cooperated with ROKID to complete the scanning payment through the glasses via voice recognition, thus establishing a new payment system.

As I mentioned before in relation to the Vision Pro glasses, based on new interactions such as iris recognition, eye tracking, and even blinking, the payment method will be replaced from camera scanning.

Now, the launch of the MCPs of WeChat and Alipay actually enables large models to have payment capabilities, and this capability can be completed based on conversations. So now developers can complete payments through AR glasses or even mobile phone conversations, provided that developers integrate biometric features, including voiceprints or the aforementioned eye - tracking technology.

The following picture shows the combination of the payment scenario completed by Alipay and ROKID glasses.

MCP is an extended capability library for AI.

Compared with building a knowledge base or fine - tuning a model on their own, some third - party capabilities cannot be achieved by AI model teams.

Even if a company has a payment license and can develop a payment technology path, it has to consider whether there are enough users and a developer ecosystem for payment. For example, on payment platforms like WeChat, users hardly need to worry about fund security or payment operations and can complete transactions quickly. At the same time, developers have a strong willingness to connect.

Although OpenAI proposed to build an e - commerce platform some time ago, because it has a product entry for the AI model, user traffic, and user conversation scenarios, it can indeed build an e - commerce system. However, building its own payment system and the logistics system required for e - commerce always takes time and user acceptance, and even merchants need to be willing to use its payment channels.

It should be noted that WeChat Pay, Alipay, and UnionPay all had to offer a lot of promotional activities to get merchants to choose their services. Even now, many banking institutions have their own payment entrances, but because they are not pure Internet platforms, only their loyal users will use them. Other users who have not opened an account with them will not use their payment methods.

Directions for product managers: The explosion of AI agent charging

This is easiest to understand by analogy with the current knowledge - payment model. In the past, users could obtain paid content from bloggers by subscribing with a fee. This is the business model of platforms like Zhihu and Knowledge Planet.

These knowledge - payment platforms complete the construction of an intermediate charging channel, and a relationship of content consumption and production is established between users and knowledge providers.

Therefore, the first thing product managers can do is to develop a business model for AI agent charging. In addition to knowledge payment, there are also obvious demands for consultation and knowledge payment in fields such as financial management and even healthcare.

In the past, building an agent involved not only the cost of paid AI computing power but also R & D and operation costs, making it very difficult to realize monetization. Apart from financing and one - time subscriptions or private deployment fees from B - end or C - end users, it was almost impossible to achieve commercial monetization.

Currently, some AI agent products with traffic realize monetization through advertising. However, at the conversation level, users can only make a one - time purchase or subscription, and the commercial conversion rate is very low. Especially now, most paying users have chosen leading models like GPT or Claude and are not willing to pay for a standardized product anymore.

However, after the paid WeChat MCP is launched, the entrance of mini - programs and conversations can obviously increase the user traffic entrance. According to the pirate model AARRR, during the conversation process, users may be required to pay for certain topics, tasks, or questions. As long as it is not involved, it is free.

This greatly increases the stickiness and activity between users and the product.

For example, the following is a case of WeChat MCP.

The above is a case of the payment scenario of WeChat MCP, which is also shown in the official documentation. Currently, it is marked as an experimental version because its refund and appeal mechanisms are not yet perfect. That is, users can pay for agents, but they cannot get a refund or make an appeal for now.

Currently, Alipay supports all user operations such as payment, query, and refund, while the MCP of WeChat Pay is relatively weak. However, I believe that the refund and order query functions will be launched within this year.

The second direction for product managers: A new AI - paid business architecture

In addition to knowledge payment, the agent MCP can also lead to a new e - commerce architecture for AI - native products. Users can choose to buy products such as medicines and toys based on the answers obtained from the AI model. Only with a payment tool can an e - commerce platform rise.

When Alibaba first emerged, because banks did not support its payment method, Jack Ma built Alipay himself, which provided platform security for buyers' payments. This finally completed the closed - loop of e - commerce and the first order was made.

Therefore, building a new AI - paid e - commerce product architecture is also a way for product managers. The following is the construction framework of the product, which I have drawn in a prototype wireframe diagram. It is a possible way.

Finally, another area of business explosion is that the MCP developer forum and product design teams can engage in knowledge payment. Because the demand from payment system developers and product managers in China is huge. I can predict that in the future, UnionPay, China Merchants Bank, even the four major state - owned banks, and even the digital RMB will launch MCPs. These payment MCPs will complete the payment channels around their own payment systems.

Therefore, the skill requirements for payment - related product managers are bound to be updated. If a product manager has not worked on an AI model, they will surely be unable to discuss MCPs.

That's all for my sharing.

This article is from the WeChat official account "Kevin's Little Bits of Changing the World" (ID: Kevingbsjddd), written by "Kevin's Stories". It is published by 36Kr with permission.