Alipay recently launched an AI application called "ZhiXiaoBao". According to the Apple App store, the application was launched on August 28 with version 1.0.0 and updated on September 4. On September 5, at the 2024 Inclusion·Bund Summit, Ant Group officially announced the launch of "ZhiXiaoBao".

It is reported that based on Ant Bailing's large model, ZhiXiaoBao is the first domestic service-oriented AI independent app that can connect to the Alipay ecosystem and has a scene perception system to intelligently recommend scene-specific services. The ZhiXiaoBao team stated: "Our (ZhiXiaoBao) basic ecological construction logic is based on the ecosystem of 'Alipay', so when Jin Jie launched the ecosystem, it was also a corresponding 'upgrade' of the Alipay agent ecosystem, evolving from connecting many mini-programs to the logic of agent connections in the future."

In summary, the logic of AI is used to reconnect the Alipay ecosystem, and Alipay and ZhiXiaoBao can now be considered as two parallel applications within the same ecosystem. This ecosystem is centered around payment, lifestyle, financial management, and more, and from now on, when discussing the name "Alipay," it is necessary to distinguish between the Alipay ecosystem and the Alipay app itself.

The idea of using AI to connect service ecosystems is Agent (agency) AI, especially lifestyle AI agents, which have always been a popular direction for AI commercialization. Logically speaking, the Alipay ecosystem is already mature, and AI is not a new concept, so why are lifestyle AI agents emerging only now?

Based on the development of AI in recent years, it took more than a year for ZhiXiaoBao to be refined, and Ant finally found a position that maximizes the development of the Alipay ecosystem; Meituan, another lifestyle service platform, had also launched a similar product called "WenXiaoDai" before ZhiXiaoBao, and now the launch of ZhiXiaoBao brings new variables to the development of lifestyle service platforms.

01. Why Now for ZhiXiaoBao

ZhiXiaoBao has a relatively clear evolutionary path.

It is generally believed that the predecessor of "ZhiXiaoBao" was an AI assistant project named "Anna" initiated internally by Ant Group in 2018, but this was too long ago to determine the direct impact of this project on "ZhiXiaoBao." It was not until 2023, after the explosion of GPT, that Ant's self-developed "Bailing Large Model" was completed, marking the true beginning of the groundwork for "ZhiXiaoBao."

In other words, the centripetal force brought about by the concept of the large model has shown the potential of AI in the lifestyle service ecosystem.

As early as a year ago (September 2023) at the Ant Inclusion Conference, when Ant announced its financial large model, it also announced that the internal beta test of "ZhiXiaoBao 2.0" had been underway for nearly six months, with application directions involving financial consultants, insurance agents, investment research, financial marketing, insurance claims, and more. Until May of this year, "ZhiXiaoBao" was still positioned by Ant as an AI focused on "financial assistants."

In April of this year, Alipay began gray-scale testing a new AI Agent for handling affairs on its homepage. Users can find this AI assistant by continuously swiping down on the Alipay homepage, which is the predecessor of the current ZhiXiaoBao. According to official information at the time, this Alipay intelligent assistant and financial assistant "ZhiXiaoBao" were actually listed as different products by Ant Group.

Now, the intelligent assistant that was once embedded in Alipay has adopted the name "ZhiXiaoBao" and become an independent lifestyle AI service app officially available to the broader C-end market, while Ant Group has relaunched a financial service AI called "MaXiaoCai," which is already part of the Alipay ecosystem.

It can be seen that after multiple iterations, the birth process of ZhiXiaoBao was also tortuous, especially since the explosion of GPT until now, when the buzz around generative AI has always been dominant. Therefore, neither ZhiXiaoBao nor lifestyle service intelligent assistants attracted much attention before.

However, as previously mentioned in the article "AI's Ecological Evolution: The End of Modular Integration and the Birth of a New Business Logic" in the "New Position," compared to creating a new super app, reconnecting all functions with AI technology within an already well-established super ecosystem may be one of the core narratives of the upcoming AI era for To C.

Currently, Alipay covers more than 8,000 services across over 150 service scenarios, with 700 million financial users, 600 million healthcare users, and 500 million travel users, making it a gathering place for multiple vertical groups. Accurately targeting and addressing the personalized needs of such a vast user base is precisely where artificial intelligence excels.

Alipay is inherently an internet transportation hub, but the actual development rate of its vast service ecosystem may be only 10%. According to Alipay's latest commercial vitality report, the app is not short on service-oriented traffic, with diverse service scenarios and a large-scale user base of 900 million high-net-worth members.

What it needs is an intelligent hub to activate various scenarios, enabling users attracted to Alipay by specific services or content to access a vast array of lifestyle services through the built-in intelligent assistant, achieving a "call-to-action" response.

Therefore, as external attention to agency AI increases, and as Alipay itself needs such an AI to inject new vitality into its entire ecosystem, the current ZhiXiaoBao emerges at the right time.

02. AI Agent: The Next Battleground for Lifestyle Services?

The predecessor of "ZhiXiaoBao" was already visible on the pull-down page of Alipay, and similarly, the practice of embedding AI within lifestyle apps has been attempted before by Meituan.

In April of this year, Meituan began testing an AI assistant service called "WenXiaoDai" for C-end users on a small scale. Based on the Raccoon model, WenXiaoDai is an intelligent food delivery assistant that recommends food delivery options and dining suggestions to users. However, according to screenshots shared by some social media users, WenXiaoDai can only generate textual suggestions and list related product links at the bottom of the page.

This is still a far cry from a fully functional agent capable of placing and paying for orders (as is the case with ZhiXiaoBao). Moreover, after the beta test, there has been no new news about Meituan's WenXiaoDai. Nonetheless, the birth of ZhiXiaoBao should inspire new ideas for platforms like Meituan and Douyin, which also have lifestyle service ecosystems.

Regarding the "AI Agent + Lifestyle Services" sector, among the major relevant platforms, perhaps only Meituan's "WenXiaoDai" and Alipay's "ZhiXiaoBao" can be considered relevant at present. Other content platforms like Kuaishou and Douyin have not yet fully established their own lifestyle service systems, and the biggest pain point for users of AI agents is the diversity, completeness, and interconnectivity of services within the ecosystem.

A typical example of AI Agent + Smart Home has already seen relevant products landing, such as central controls for smart home systems or smart speakers that can be considered AI Agent lifestyle service products. However, as previously mentioned in "New Position," the biggest challenge in the smart home industry is not the technology itself but rather the poor interconnectivity between smart home appliances from different brands, making it difficult to control multiple brands through a single system. Therefore, consumers with high interconnectivity requirements may choose to purchase smart home appliances from the same brand throughout their entire home.

Furthermore, the recent legal dispute between OpenAI and The New York Times underscores the difficulty of achieving interconnectivity between OpenAI and mature content and tool ecosystems. Therefore, a rich and complete ecosystem is the core moat for lifestyle service platforms to develop AI agents.

If ZhiXiaoBao's path proves successful, it is not difficult to speculate that Meituan may soon follow suit and relaunch "WenXiaoDai." In other words, the lifestyle services sector is likely to open up a new battlefield - AI agents.

The next tipping point for AI agents will be when AI truly completes the order and payment process on behalf of users. This moment may arrive soon or may encounter some setbacks. In fact, Alipay has laid the groundwork for payment technology in the past, with features like password-free payment, facial recognition payment, and palm print payment. Building upon this foundation, using AI to complete the order and payment function technically poses little difficulty. What truly takes time is cultivating user acceptance and improving relevant laws and regulations.

After all, allowing AI to spend money on behalf of humans requires caution.

03. Final Thoughts

Generative AI and Agent AI (AI Agents) are two distinct directions in the field of artificial intelligence. The former focuses on generating new, original content such as text, images, and audio by learning from existing datasets, while the latter emphasizes simulating intelligent behavior, interacting with the environment, and making decisions based on collected data to perform tasks. While they can complement each other in some aspects, they are not simply parallel and complementary. For example, the current ZhiXiaoBao is both an Agent AI and possesses generative capabilities, reflecting Alipay's previous layout in both content and AI.

However, at this moment, when discussing AI agents, the focus is no longer on the large model technology behind them but rather on the application scenarios themselves.

Based on personal testing, ZhiXiaoBao currently only stays at the stage of being able to query relevant service information. When users give instructions, ZhiXiaoBao retrieves relevant service card mini-programs, and users can directly access these mini-programs on Alipay through ZhiXiaoBao. In terms of user experience, it does save some time.

However, as mentioned above, to reach the true tipping point, this AI still needs to do more in one fell swoop.

*The cover image and illustrations in the article are sourced from the internet.