To The Moon

The AI wave has brought about several rounds of fervent online requests for beta access codes, and this time, the spotlight is on the ubiquitous super-app Alipay.

On June 15th, a tech media outlet published a report titled "Decoding the AI-Powered Alipay, the Biggest Overhaul in 20 Years is About to Launch." The following day, Alipay officially announced the commencement of an invite-only test for its AI version, releasing 100 invitation codes. Each user who gains access can then invite at least three more users, employing a classic social growth strategy that signals Alipay's commitment to what feels like a second entrepreneurial venture.

Initial Impressions of the AI Alipay

The hype is significant, and codes are scarce, but how much of this "biggest overhaul in 20 years" addresses genuine user needs, and how much is just AI posturing? A hands-on evaluation suggests the AI-powered Alipay currently feels more like an unfinished product—somewhat rough with numerous issues. However, Alipay is arguably the most suitable product within the Alibaba ecosystem for an AI-driven transformation.

A Confused Accountant, A Precise Triage Desk

Judging from the beta version, Alipay's AI strategy is not overly aggressive. The width of the original five fixed tabs at the bottom has been compressed, making room for a circular AI entry point in the lower-left corner, featuring the avatar of the AI persona "A Bao." A Bao is depicted as a female character, distinguished from the "Dou Bao" persona primarily by her long hair tied in a braid, whereas Dou Bao has shoulder-length hair. It's worth noting that Ant Group's AI assistants seem to favor names starting with "A"—for instance, the health assistant launched in late 2025 is called "A Fu." However, A Fu exists as a standalone app, while A Bao is integrated directly into the Alipay product.

Clicking on A Bao's avatar switches the interface to the AI version, defaulting to a chat dialog with her. Thanks to the efforts of products like Dou Bao, DeepSeek, and Qwen, conversational interfaces have become the most familiar mode of AI interaction for general users. Beyond hairstyle, another quick way to distinguish A Bao's interface is the presence of four permanent shortcuts at the bottom of her chat screen: Scan, Pay/Receive, Travel, and Finance—all high-frequency functions within Alipay.

Interacting with A Bao primarily enables two types of functions. The first is analyzing the spending, income, and investment data from one's Alipay account. For example, asking her "How much did I spend on Taobao in 2026?" For such requests, A Bao has some data access and can perform analysis, presenting conclusions with text and charts. Unfortunately, her calculations are currently unreliable. Like all AI assistants, A Bao suffers from significant hallucination issues.

During an initial calculation, A Bao reported that total spending on Taobao-affiliated platforms (including Taobao, Tmall, Taobao Flash Sales, Xianyu, Hema, etc.) in 2026 was 2,760.23 yuan, even providing a pie chart and noting that the proportion spent on dining was excessively high. The analysis seemed thorough, but the data was severely incomplete. Actual cumulative spending at Hema in 2026 was 386.93 yuan, yet A Bao's statistic was only 110 yuan, omitting 71.58% of the expenditure. When challenged, A Bao promptly apologized, promised a more rigorous recalculation, and then provided an even more absurd spending summary. In A Bao's world, it seemed as if Sam's Club, Xiaohongshu, and Swan Home Services had all been acquired by Alibaba and included in the Taobao platform spending tally.

As discussed in a previous article, making mistakes and apologizing is becoming a systematic strategy for AI assistants. This holds true even in Alipay's AI context, which involves financial data. After hallucinating, the AI's apology comes just as readily. However, shouldn't an AI handling financial services be more precise?

Navigating Services with AI

A Bao appears more adept at the second type of task: initiating corresponding service pages based on user requests in the dialog. Typing "I want to order takeout" brings up the Taobao Flash Sales page. "Beijing medical appointment" opens the Healthcare page. "Hail a ride to Sanlitun" triggers a Gaode ride-hailing card. Asking "Please turn the money in my Alipay into 1 million" elicits a response that the request is beyond her capabilities, humorously adding that if she could conjure wealth with a finger snap, she'd be traveling the world instead of chatting here.

In essence, beyond summarizing information and casual conversation, A Bao's ability to provide services resembles a triage desk at a large hospital. She can navigate users to existing Alipay services with one click, but cannot magically create services that don't already exist. This restraint, however, is not necessarily a bad thing.

Financial Prudence in AI Services

The AI-powered Alipay shows relative restraint in financial services, avoiding readily recommending specific funds or financial products to users. While users can ask the AI for thoughts on preserving and growing their assets, Alipay's approach is limited. It can access your account balance—revealing one's financial status to an AI for the first time—and suggest allocation ratios, such as what portion to keep in Yu'ebao, invest in fixed-amount plans, or allocate to funds. However, no matter how the user inquires, Alipay remains tight-lipped, never recommending specific funds. It only provides direct links to channels like fund pages, leaving the final decision on whether and what to buy entirely to the user.

Capable of Much, Achieving Little So Far

After 22 years of feature accumulation, today's Alipay has become quite "bloated." According to QuestMobile data, the scale of Alipay mini-programs has exceeded 4 million, with monthly active users reaching 644 million in March 2026. Analysis of the top 100 mini-programs shows that lifestyle services are the core use case, accounting for 47%.

Current AI capabilities are primarily applied in two scenarios: productivity and lifestyle services. Products like CodeX, Trae, Kling, Coze, and WorkBuddy target the productivity scene, characterized by concentrated user scenarios and focused, deep vertical capabilities. The AI-powered Alipay targets the lifestyle services scene: a concentrated core surrounded by a vast, fragmented array of personalized scenarios radiating outward.

Alipay's initial core scenario was payments, gradually expanding to include phone top-ups, appointment booking, bill payments, social security queries, wealth management, insurance, and public transit. Similarly, WeChat's core was messaging, expanding into red packets, payments, and the same suite of peripheral services. Indeed, the latter seven items are copy-pasted, highlighting the clear convergence and even encroachment on each other's core services as these mega-apps expand their scenes.

Today, both Alipay and WeChat host a massive number of mini-programs, covering nearly every aspect of daily life. This very "bloat" provides a convenient foundation for their AI transformations. Many AI products struggle with having capabilities but no clear tasks, forcing them to seek out problems to solve. For Alipay and WeChat, the potential tasks are already abundant.

The Core Challenge of Interaction Design

When a product aggregates too many disparate functions, enabling users to quickly find what they need becomes the central challenge of interaction design. Before AI, the traditional logic was to bury less-used features deeper in the interface. In the era of "conversation as a service," an AI assistant can theoretically access all services within its ecosystem with equal efficiency. A niche function used by 10,000 people should be as easily accessible as a mainstream function used by 10 million daily.

However, from another perspective, adding AI to an already feature-complete ecosystem like Alipay may not deliver a dramatically upgraded user experience. It essentially swaps a "search + click" interaction for a "conversation + click" one. The difference between "say a sentence and it hails a ride for you" and "say a sentence and it opens the ride-hailing page for you to continue yourself" is akin to the gap between receiving a Mercedes-Benz car and a 100-yuan Mercedes-Benz discount coupon.

With unreliable accounting and service linking that offers mostly superficial convenience, the current AI-powered Alipay is undoubtedly a work in progress. Recently, Dou Bao launched a product recommendation feature, allowing users to go directly to Douyin's marketplace via recommended links. Even if this doesn't become the primary shopping path, Dou Bao's massive user base could drive significant e-commerce traffic through this new funnel.

A Defensive Transformation

Compared to an aggressive charge, Alipay's AI transformation seems more defensive. It aims to solidify its own service ecosystem before Dou Bao and DeepSeek can link up more e-commerce and local lifestyle service capabilities. If the era of "conversation as a service" fails to materialize, A Bao can quietly recede, returning the home screen space to the compressed bottom tabs. But if that era arrives swiftly, and everyone starts wanting that "Mercedes," users will at least remember they once claimed a 100-yuan discount coupon for it.