Alipay does not want to be a supporting player in the AI era

Author | Hanyu Wang

Editor | Fan Zhang

Bring up the AI interface on Alipay's home page, where the dialog box replaces the densely packed mini-programs; users say "Find nearby milk tea coupons" to "Abao", and the promotions of surrounding stores are automatically matched, with verification and order completion done in one step.

Recently, Alipay completed its largest redesign in its 22-year history of launch.

Earlier this month, the AI-powered version of Alipay, "Abao", officially launched full public beta testing. Almost simultaneously, WeChat Pay's "AI Exclusive Card" was also implemented within the WorkBuddy agent ecosystem.

Since entering 2026, leading platforms and various payment institutions have intensively unveiled their AI strategies, quietly kicking off the agent competition in the payment industry.

From "Leave Immediately After Use" to "Initiate Conversations Proactively"

Along with the AI transformation process, payment apps have for the first time seen user behaviors that go beyond their traditional tool-oriented attributes.

After the AI version entered internal testing, an Alipay representative told 36Kr that one phenomenon surprised the team: many users opened Abao not to complete specific tasks, but simply to "play around" with it — such as casual chatting, asking imaginative questions, and testing various responses. "'Abao' feels more human-like, which has sparked users' interest in interacting with it. We were quite surprised that many users come to 'play around' with Abao. This kind of behavior would have been almost impossible on the traditional version of Alipay."

This is a landmark signal.

Over the past decade, the core pursuit of payment apps has been "efficient, seamless, and leave immediately after use" — the shorter the user's stay time, the better the experience was considered.

Alipay, for example, was previously a payment tool that existed to support specific scenarios, with naturally limited user stickiness. Especially in recent years, as new tracks like value-for-money e-commerce, content-driven e-commerce, and local retail have continuously diverted traffic, the flow of traditional shelf e-commerce has been further fragmented. Even as an industry leader, Alipay has had to frequently use red envelopes, promotional subsidies, and other measures to boost user activity.

AI is overturning this logic, pushing payment tools from behind the scenes to the forefront.

In the past, to buy a cup of milk tea, users typically had to open a food delivery app to place an order first, then actively choose Alipay to complete the payment — payment was just the final step of the transaction closed loop. Now, by stating their needs in the "Abao" dialog box, users can complete product selection, coupon redemption, order placement, and payment in a one-stop process, making "Abao" a new entry point for consumption decision-making.

In short, payment apps have gained "entry-level" user stickiness.

From a broader perspective, the entire payment industry is currently undergoing a similar interaction paradigm shift: moving from "users seeking services" to "services proactively reaching users". The industry's general consensus is that AI will not replace the existing payment process, but will add a layer of intelligent decision-making capability at the application level, ultimately changing how users engage with payment services.

Alipay has chosen the most thorough path: a product-level reconstruction.

"Alipay's AI strategy is a choice oriented toward the future of the agent economy. We believe that the future human-service interaction will be restructured, the current intermediate carriers based on apps will be replaced by agents, and this requires a new architecture to adapt," an Alipay representative stated.

According to 36Kr's understanding, Alipay's AI exploration started in the second half of 2023. It initially began with an on-device intelligent voice assistant, gradually launched secondary agents for vertical fields such as government services, transportation, and employment, and only launched the end-side reconstruction "Bao Project" after accumulating sufficient data.

In terms of user experience, the reconstructed AI-powered Alipay essentially folds tens of thousands of services that were previously hidden under multiple layers of menus into a single dialog box. It has currently launched 72 high-frequency service capabilities, covering core scenarios including account inquiries, bill payments, government services, and transportation.

What further demonstrates its strategic direction is that this architecture is open to the entire industry. Not long ago, Alipay launched its AI open platform, allowing third-party services that meet access standards to connect to "Abao" via the MCP/skill protocol and provide services to its 1 billion users.

WeChat Pay has taken a different path.

Instead of building its own centralized AI entry point, it has launched the "AI Exclusive Card". This is defined as an atomic-level payment capability embedded into the service workflows of third-party agents. When users generate consumption demands within an agent, the AI selects products and initiates payment, which is ultimately completed via the WeChat Pay AI Exclusive Card.

A WeChat Pay representative told 36Kr that the design goal of the AI Exclusive Card is very clear: to make agent payments secure and controllable, and make them accessible for third-party agent platforms to call. "Which agent the user generates a consumption demand in, how the agent understands the demand, what products it recommends, and what interaction it designs — these are the unique values of each agent platform. WeChat Pay only provides a set of capabilities within the scope of user authorization, ensuring dedicated funds for specific purposes and confirmation for every transaction, at the final payment stage."

One player is doing product-level reconstruction, while the other is offering standardized outputs. Strictly speaking, the AI-powered Alipay and WeChat Pay's AI Exclusive Card are not in the same dimension, but in terms of changes to the payment process, the paths of these two leading players both point to the same trend: pushing payment from a supporting role to a more prominent position closer to order placement decisions.

All Players Unveil Their AI Strategies

The trend of shifting from passive to active is similar, but the implementation paths of different payment tools have their own focuses, which are rooted in the inherent differences in each party's data assets and ecosystem structure.

Over its 22 years of operation, Alipay has deeply integrated service and transaction data. According to its official data, Alipay's ecosystem hosts millions of mini-programs and over 8,000 types of services — ranging from wealth management and insurance to utility bill payments, government service processing, and travel ticket booking — a large number of services are natively built within the Alipay system. This means the supply of services that AI can directly schedule is sufficiently robust.

The offline network is another major advantage. On the 8th of this month, Alipay announced that its "Tap to Pay" feature has completed an AI upgrade, transforming 30 million offline touchpoints into an AI operation network, turning millions of merchant devices into "Tap Device Agents". Users can tap to access services such as consumption coupons, membership benefits, and marketing activities scheduled by the merchant agent "Xiaoyu", which is equivalent to laying an AI entry network across the physical offline world.

With "Abao" online and "Tap to Pay" offline, Alipay's AI landscape presents a dual-entry structure — the online conversation entry handles digital services, while the offline touchpoint network serves physical commerce, with the two connected via a unified open AI infrastructure.

WeChat Pay's strength lies in the breadth of its ecosystem. The WeChat system is home to a massive number of official accounts, mini-programs, video accounts, and third-party agents, where consumption demands are scattered across various scenarios. Under such an ecosystem structure, making payment capabilities standardized and opening them up to allow every agent to connect securely is the most efficient choice.

JD.com launched its A2P2 agent autonomous payment protocol in June this year, leveraging the depth of its e-commerce and supply chain scenarios to provide a phased implementation framework for different use cases. Its AI payment capabilities cover both C-end voice payments and B-end supply chain automatic settlement, with a relatively fast implementation pace in enterprise-level scenarios.

Different platforms choosing different paths essentially reflects their respective resource endowments. Players with their own service ecosystems tend to build entry points, those with scattered scenarios focus on enabling connections, and those with strong supply chain foundations prioritize B-end implementation. The coexistence of diverse paths forms the current overall landscape of AI transformation in the payment industry.

Trust Is the First Threshold

Regardless of the path taken, AI payment cannot avoid the same question: Will users dare to entrust their money to AI?

This is not a technical issue, but a trust issue. The underlying logic of current industry solutions is very similar — retaining the final decision-making power in the hands of users.

WeChat Pay's approach uses physical isolation plus transaction-by-transaction confirmation. The "AI Exclusive Card" is set up within WeChat's balance wallet, where users can transfer funds on their own, completely isolated from their main account. For each transaction, the AI is only responsible for product selection, order placement, and initiation, while the final deduction must be reconfirmed by the user with their password; users can transfer funds in and out of the card at any time.

"The premise for users to dare to entrust their money to AI is that users always know, and can always control how much the AI can spend and where it spends," a WeChat Pay representative noted.

Alipay has built a systematic security and trust infrastructure, including mechanisms such as Blue Shield identity verification, full-process transaction evidence preservation, and consumption limit authorization control, to ensure that the risks of unauthorized charges and misuse in cross-device AI automatic payment scenarios are controllable, with all agent operations and payment behaviors traceable.

Restraint is the common choice across the entire industry. In this field with strong security attributes, building trust is far more important than improving efficiency.

Looking back at the industry's development, mobile payment spent a decade completing the efficiency revolution of "moving cash into mobile phones", and AI is expected to bring an essential leap in payment's role — from a supporting tool to a leading entry point. When the payment process is deeply integrated with AI decision-making, payment institutions will no longer only hold transaction flows, but also the pre-decision entry point for consumption.

The AI battle has just begun, and it remains to be seen which players on different paths will ultimately succeed. But one thing is certain: when payment transforms from the end of a transaction to the starting point of services, the commercial value this industry carries will far exceed the revenue from transaction fees itself.

*Disclaimer:

The content in this article only represents the author's personal views.

Investing involves risks, and caution is advised when making investment decisions. Under no circumstances shall the information or opinions expressed in this article constitute investment advice to any person. Before making any investment decision, investors must consult professional advisors and make prudent decisions if necessary. We do not intend to provide underwriting services for transaction parties or any services that require specific qualifications or licenses to conduct.

Follow for more updates