If Abao is born, why does Afu need to exist?

In April, I had a minor surgery. Just then, Ant had recently launched Afu, and there were ads everywhere. So, it was natural for me to become a user of Afu. I'm quite a loyal user, chatting with it several times a day.

I've sincerely recommended Afu to many friends. I think it's much better than my attending doctor, who is difficult to make an appointment with and stingy with words. But at the same time, I had a doubt -

Ant spent a lot of money on advertising to attract users for Afu, but aren't the users already on Alipay?

Of course, at that time, I understood such an arrangement. Because the former is a digital - life tool, and the latter is a health AI product, it makes sense to acquire customers separately.

But this explanation didn't solve my doubt. It only answered "why separate them", not "how to connect them".

Actually, the core problem of Alipay is not "attracting external users", but completing a match on its own platform: connecting users with services. Just like what Tinder did in 2012.

When Tinder invented the right - swipe in 2012, the only core problem it solved was to make the successful matching simple and exciting.

So, when Ant launched the AI - version of Alipay on June 16th, with a right - swipe, Abao appeared, and all kinds of services were folded into one dialog box. At this moment, Alipay finally achieved a successful match.

But matching is just the beginning.

Two people match successfully online, but to become a long - term offline couple, besides a good first impression, they also need to see if they can meet each other's functional needs. It's very practical and complex.

In a way, for Abao to truly become users' AI entrance, it also needs to solve the functional problems of a partner. Can you make money, do work, or be beneficial to physical and mental health?

You have to give users a reason to swipe right. Otherwise, they won't pay more cost for a software, even if it only takes a movement of the index finger.

I think Alipay has verified this conclusion in the past two decades. Without sufficient reasons, it's difficult for users to stay.

So, let Abao absorb Afu, Lingguang, Maxiaocai, etc. Because the battle for the AI entrance has actually entered the middle stage, and the time window is very narrow.

01

The competition in AI to C is different from all the Internet competitions people are familiar with in the past.

In the era of traditional apps, users can install more than a dozen tool - type apps at the same time. They use Gaode today and Baidu Map tomorrow, with different divisions of labor, and user habits can be cultivated gradually.

But it's not the same for AI assistants. The core value of an AI assistant is based on its in - depth understanding of you, such as your consumption preferences, health status, asset structure, and historical commands. These data can only be truly accumulated when you use the same assistant for a long - term, high - frequency, and continuous period, and then the AI can really "understand you".

This means that the competition in AI to C naturally tends to be a winner - takes - all situation. Users will ultimately choose a main entrance and be deeply locked in.

Just like in the era of search engines, Google defeated Yahoo, Lycos, and AltaVista. It didn't rely on technological superiority but became users' "first reaction" first.

Once users' habits are solidified, it's almost impossible for latecomers to turn the tables.

Goldman Sachs defined 2026 as the "critical year" for Chinese Internet giants in its latest research report. In 2026, it's not a "model competition" but a life - and - death battle for the "default entrance". Whoever gets the default entrance will rewrite the traffic distribution, advertising budget, transaction closed - loop, and even the entire Internet profit pool.

According to the data from QuestMobile in the first quarter of this year, Doubao's monthly active users have reached 345 million, with a gap of 180 million from the second - ranked app. More notably, ByteDance's products occupy 7 positions in the top 20 of monthly active users in AI applications. Doubao is the super - entrance, while MengAI, Doubao Aixue, and Maoxiang are positioned in vertical fields, forming an internal cycle in the entire ecosystem. Users are getting more and more involved in ByteDance's AI world.

This pattern is not completely solidified yet, but the solidification speed is faster than expected. The time left for Ant may be shorter than it thinks.

02

Ant actually has a good hand of cards.

Alipay has one billion monthly active users, which is a natural user base for Abao. This is a starting point that any AI startup company can't replicate in the next decade. Ant has deep - seated data accumulated over the years in financial and medical scenarios, such as medical insurance information, credit systems, user asset structures, and risk - control models. These are what Doubao doesn't have and what WeChat can't get in the short term.

But this hand of cards is now being played in four parts.

Afu holds health data and consultation scenarios. Its monthly active users have exceeded 100 million, making it the world's largest health AI application. Maxiaocai holds wealth - management users and risk - control capabilities, and has sent more than 120 million rational investment reminders to users. Lingguang holds the technical ability of "generative applications". Although its monthly active users are only 3.47 million and far behind, its exploration in multi - modal interaction has accumulated valuable engineering experience.

These three things are exactly what Abao needs the most.

When a user opens Alipay and tells Abao "I've been having some heart discomfort recently", Abao should be able to directly retrieve his historical health records, help him make an appointment with a suitable specialist, and complete the entire medical insurance payment process. This is what Afu is doing, but it's locked in a separate app, and there is only a shallow functional call between it and Alipay, without real data integration.

Similarly, when a user says "My income has increased recently. Help me find a wealth - management plan", Abao should be able to retrieve his asset structure, give suggestions based on the current market situation, and complete the operation directly in Alipay. This is what Maxiaocai is doing, but it also exists in the form of an independent app, with two sets of user - operation logics and two sets of data warehouses.

What Abao can do now is be a dispatcher. But what Abao should do is be a real butler who understands you.

Besides functions, the business imagination also calls for integration. As we all know, it's difficult for Chinese To C Internet products to charge through subscriptions. Recently, Doubao has started to try, but judging from the market feedback, the resistance should be great.

So in the To C market, it's always the case that the cost is borne by others. I can provide free services, but please watch the ads. In the past few years, Alipay has started to try the short - video business. Isn't it trying to find a new growth point in the advertising and marketing business?

I even think that the complex design of the platform's home - page functions, including Alipay, is to some extent to promote ad exposure, just like a complex layout in a shopping mall increases the probability of consumers passing by stores.

But the AI entrance war has pushed the business logic to the other side. The entrance has become a "toll - booth on the highway", and the user's path will become very simple.

This model will even affect the original ad exposure. Once users swipe right, the short - videos will be out of their sight.

So for Ant, it must find a new monetization path, with more "lanes" leading to different destinations and attracting more "cars" to enter through the entrance.

Asking about today's weather or how to cook braised chicken legs are not "traffic flows" with commercial value. The "traffic flows" that can really make the platform earn "toll fees" have clear destinations, such as where to see a doctor, what funds to buy, or buying tokens for vibe coding.

Ant has paid a lot of infrastructure costs to build these branches. Take the health business as an example. In early 2025, Ant acquired Haodf.com.

Even as an ordinary user, I can feel the changes in Haodf.com over the past year. It has become very aggressive.

Not long ago, I took my child to Beijing Children's Hospital. After leaving the consulting room, the expert's assistant chased after me and insisted that I scan the doctor's Haodf.com QR code.

But it's not just about scanning the code. I was also required to repeat the entire consultation process online. To be honest, my first impression at that time was "a waste of effort".

But after I got home and thought about it, the whole process actually digitized the offline medical records, enriching Haodf.com's data assets, which may be used to train Ant's health model.

Doctors and hospitals surely won't cooperate with this process for no reason. There must be an incentive mechanism behind it. It seems that Ant is continuously paying costs for the health business. For a commercial institution, it needs more imaginative income to plan its returns.

03

Of course, I also know that integration is difficult.

The resistance may not come from technology but from the organization.

Whether it's Afu, Maxiaocai, or even Lingguang, they are all managed by specialized teams. Each team has its own structure, performance evaluation, and budget. If integration is to be carried out, it is necessary to connect, centralize, and even merge similar items. For the organization, the integration process requires a cost of game - playing.

The loss - aversion theory of behavioral economist Kahneman says that people's aversion to loss is twice their desire for the same amount of gain. When this rule is magnified to the organizational level, the effect will be even more significant. Behind each product to be integrated, there is a team that doesn't want to "disappear" and a research and development investment that no one wants to admit as a sunk cost.

Microsoft is a valuable reference case. It took Nadella nearly a decade to integrate the scattered AI capabilities of Cortana, Bing AI, Azure AI, Teams AI, and Office AI into the Copilot brand.

During the integration process, some teams were laid off, some leaders left, and some product lines were forcibly closed. These costs are real losses in the short term, but after the integration is completed, Microsoft's competitiveness in the AI era is truly revealed.

Nadella once said that the most difficult part of doing this is not the technology but convincing the internal team to believe that "the whole is greater than the sum of the parts".

Of course, integration doesn't necessarily mean that only Abao will survive in the future. The logic can also be like this: Afu doesn't disappear, but it becomes Abao's "health - service layer", and the user entrance is unified to Alipay, with the Afu brand retreating to the functional level. Similarly, Maxiaocai becomes Abao's "wealth - management layer", and users no longer need to download a separate app. Lingguang's flash - application ability can be integrated into Abao as a content - generation function, so that the AI assistant in Alipay can also help users "generate a book - keeping tool".

Only after integration can Abao truly be qualified to compete with Doubao and WeChat AI head - on, not relying on just a dialog box but on a real AI butler that can integrate health, wealth, payment, and life.

The value of a super - entrance is far greater than that of several functional apps. This has been repeatedly proven in the history of the Internet. Fighting separately often means being defeated one by one.

Currently, Afu, Maxiaocai, and Lingguang still have their own user groups and data assets, and they haven't been severely eroded by competitors. This is the best time for integration - proactive integration can transform the accumulations of the three product lines into Abao's capabilities, achieving an effect of 1 + 1+1 > 3.

But if Ant continues to wait, the cost of future integration will be higher, and once the competitors also take action, the competition cost will also increase. Once the WeChat AI assistant runs smoothly, Afu's health users will find that "WeChat can also provide medical consultations", and the success of Doubao Aixue has proven that ByteDance's penetration ability in vertical scenarios cannot be underestimated.

Alibaba has provided a reference. After several years of practice and exploration, Alibaba upgraded Tongyi App to Qianwen, merged the Tmall Genie team into Quark, and connected more than 400 capabilities of the Alibaba ecosystem to Qianwen at once, declaring with actions that there is only one main entrance for C - end AI.

This process is not smooth, and there must be internal resistance, but Alibaba chose to exchange short - term pain for long - term competitiveness.

As a result, Qianwen's monthly active users reached 166 million in the first quarter, with a month - on - month growth of 969%, becoming the fastest - growing AI application in China. This achievement is related to the integration.

The success of Tinder's right - swipe is not just because it designed a smart gesture. The deeper reason is that Tinder has never launched a second app called "Tinder Health Edition" or "Tinder Social Edition". The whole company has only one thing to do, only one goal, and only one moment that users will always remember.

With a right - swipe on Alipay, Abao appears. I hope this is a good start.

But a start alone is not enough.

This article is from the WeChat official account "Most Talk FunTalk" (ID: iFuntalker). Author: Wang Fangjie, Editor: Liu Yuxiang. Republished by 36Kr with authorization.