"China Muse" cannot pass Doubao's review
Recently, many users have noticed that Doubao App now supports the payment of utility bills including water, electricity and gas fees. After entering prompts like "pay water bill", Doubao will remind users to bind their account numbers and guide them to the life payment interface.
Alphaboard (ID: wujicaijing) found through actual testing that the above interface is not embedded inside Doubao, but a life service mini-program previously launched by Douyin.
A source close to Doubao said that Doubao is gradually building service capabilities for travel scenarios, utility bill payment and other daily life scenarios.
Expanding capabilities to handle personal affairs is the key investment direction of Doubao in the recent period.
Shortly before the National Day holiday, Doubao launched the "Use Doubao for Travel" feature, which gathers four functional modules: map navigation, transportation, hotel accommodation, and food and entertainment, covering functions such as navigation, nearby services, ride-hailing, high-speed rail, air tickets, hotels, homestays, catering, and sightseeing.
Earlier, Doubao also launched the "Ask Doubao Before Buying" feature, which provides product purchase suggestions and links, allowing users to complete consumption without jumping out of the App.
Services like utility bill payment, ride-hailing and hotel booking are characterized by triviality, rigid demand and personalization. By handling these tasks well, Doubao can gradually cultivate users' mindset that it "can handle personal affairs".
With the expansion of personal service capabilities, Doubao, which started as a Chatbot, is increasingly taking on the attributes of a Personal Agent.
Building a "Chinese version of Muse" is becoming the new focus of major domestic technology companies. Companies including Alibaba and Tencent have already set off, and ByteDance is also driving Doubao to accelerate its catch-up. But the real question is:
Personal agents like Muse may find no suitable application scenarios in China.
A key prerequisite for the great success of personal agents in the US market is that local internet life services are far from satisfactory, lacking all-encompassing O2O platforms and mini-program ecosystems, and the digitalization level of most merchants is very low.
Taking in-store catering as an example, US users have relied on calling restaurants to reserve seats for many years; Muse and Instinct have taken over this part of the work through multi-agent collaboration, easily hitting the pain point.
But the Chinese market is completely different — the mobile internet here is highly mature and extremely developed. Life services such as catering, travel and ticket booking have long been fully internet-based, with sufficient supply and smooth processes. The intervention of Agents to forcibly change user actions will not bring immediate improvement to the entire service process.
It can even be said that the attempt of major technology manufacturers to agentize the internet life service ecosystem is largely wishful thinking, which will not only fail to improve efficiency and experience, but also bring new challenges such as personal data security risks.
In any case, major manufacturers including ByteDance have taken the first step towards personal agents. A new AI battle is about to break out.
1
Personal agents have just started, and have gradually shown different forms in China and the United States. In terms of growth conditions, the biggest difference between personal agents in China and the United States is:
Chinese companies are backed by a sound internet infrastructure, and can make full use of online and offline service capabilities; US players started earlier, but are still working hard to open up service links and build fulfillment capabilities on their own.
For example, in the scenario of restaurant reservation, various Agent services of Chinese companies can use the A2A protocol to conveniently call the Agents of local life platforms, and smoothly complete a series of operations.
But in the United States, personal agents such as Muse and Instinct still rely heavily on "AI phone calls". Because the probability of AI calls being hung up by merchants is too high, they have also launched real-person call services, which is equivalent to equipping AI assistants with human assistants. The low efficiency and high cost are imaginable.
This also explains why when Muse gradually gains momentum and the whole industry is hyping personal agents, major domestic manufacturers have entered the market one after another, but have not followed up immediately at the product level — the functions of Muse are already achievable by major manufacturers, and the only gap is the lack of cloud computers.
For Doubao, besides being backed by the unparalleled O2O service capabilities of the Chinese internet like other companies, it also has its own unique advantage: standing on two boats.
Doubao was born in the AI era, but it has one foot firmly planted in the highly mature mobile internet ecosystem, and has been built as a super App that fits the mindset of mobile phone users since its birth.
One of the manifestations is that Doubao ranks first in its track in all the most important indicators valued by the mobile internet, including monthly active users, daily active users, and average daily usage duration. With 380 million MAU and 140 million DAU, Doubao has been one of the Apps with the largest number of users in China; its average monthly usage duration of 143.7 minutes and average number of openings per capita of 76.7 times are also envied by a large number of tool-type Apps.
With a large and active user base, Doubao can easily connect cross-platform ecosystems and quickly access various service capabilities.
For this access to water, electricity and gas bill payment functions, Doubao uses Douyin mini-programs as the receiving end. Although the ecological prosperity of Douyin mini-programs is not as good as that of WeChat and Alipay, the gap mainly lies in long-tail scenarios, and it is sufficient for mainstream daily scenarios.
At the same time, it can find a large number of partners from outside. For example, "Ask Doubao for Travel" has introduced third-party services such as Caocao Mobility, Flight Manager, High-speed Rail Manager, and Baidu Maps.
These internal and external services provide scenarios for Doubao's traffic, and also expand the service capabilities of personal agents.
In contrast, overseas services such as Muse and Instinct have seized the first-mover advantage of personal agents, but face an unsolvable problem: the O2O foundation of the US internet is too poor, and the digitalization and internet access level of small and medium-sized merchants is very low, and they all rely on making phone calls and sending emails to do business. To connect them to the service closed loop, AI has to do the clumsy work of making phone calls.
The other foot of Doubao is firmly planted in "AI-native".
It is a product that meets the expectations of users in the AI era. It can not only act as a Chatbot to chat, accompany users and provide entertainment to deliver emotional value, but also provide services in scenarios such as office work and personal affairs handling, with very obvious Agent characteristics.
Coupled with product and operation capabilities that are far stronger than its peers, Doubao's AI-native user mindset has been solidified. In the eyes of many C-end users, Doubao is almost synonymous with AI applications, and even synonymous with AI. This AI-native feature and mindset allow Doubao to achieve twice the result with half the effort in building personal agents.
Abroad, products like Muse are admittedly closer to the strictly defined personal agents: independent cloud computers, personal exclusive memory, task disassembly and cross-application execution, secondary confirmation for high-risk operations, etc.; but limited by various shortcomings of the US internet and business ecosystem, their actual work efficiency is not necessarily higher than that of Doubao.
Now, Doubao is only one step away from a complete personal agent — it has not yet equipped each user with an independent cloud computer. But this is only a matter of technology and cost. In terms of capabilities and experience, Doubao has caught up with and even surpassed its peers, but has not yet officially raised the banner of personal agents.
2
Doubao's recent accelerated progress towards personal agents implies a possibility: it is preparing to skip the disputes between AI programming and office Agents, and directly compete for the crown of the next era.
An awkward fact is that after ByteDance won the Chatbot battle, it has not won a major victory in the AI track for almost two years.
This does not mean that the number of Doubao's users has decreased. In fact, according to data from market research institutions such as QuestMobile, Doubao's leading advantage is still expanding, and its growth rate is significantly better than that of all its competitors.
But continuing to "suppress" AI Apps such as Qianwen and Yuanbao does not mean that Doubao and ByteDance AI have achieved great success. The real problem ByteDance faces is that the AI industry is developing rapidly, the trend of AI applications is constantly shifting, but Doubao has not stood at the forefront of the wave.
After Chatbot, the first major trend in the AI industry is AI programming. This field extremely tests the code capability of the model itself, which is not ByteDance's strength.
In fact, in the entire AI programming track, domestic AI companies have not become the real protagonists; although some models have topped various token call lists from time to time, they mainly win by cost performance, and Silicon Valley giants are still better in terms of absolute performance.
Doubao has also retained its position in this area. To this day, AI programming still does not occupy the main entry in the Doubao App.
After AI programming, office Agents have risen. Compared with the former, it relies more on the deep integration of AI and collaborative office platforms, which falls into the comfort zone of major domestic manufacturers, and all of them are actively participating in the competition. ByteDance not only went all out to develop Doubao Work, but also separated part of Feishu's business to Doubao.
But up to now, the situation of this battle is still unclear. Although Doubao Work has a good reputation, it cannot be said to have completely surpassed Alibaba Qianwen Office and Tencent WorkBuddy.
For ByteDance, although these two popular tracks have sufficient gold content, its advantages in traffic and users cannot be exerted, and it is easy to fall into close-quarters combat after entering the market.
The fact that ByteDance failed to win these two battles was actually predetermined from the very beginning.
Maximizing the release of the C-end advantages of Douyin is ByteDance's basic strategy for expanding its business lines. Specifically in the AI field, in the past, this advantage was mainly reflected on the model side: content platforms such as Douyin and Toutiao provide the largest corpus in the Chinese internet; especially for Douyin, the excellent performance of Seedance is inseparable from the content support of Douyin.
But providing training corpus for AI only utilizes part of Douyin's C-end advantages — the content pool, and the user pool has not been fully tapped. Douyin is of course also diverting traffic to AI applications such as Doubao, but this is obviously a rather clumsy approach, even close to wasting resources.
There may be a lot of helplessness behind this: AI programming and office Agents are oriented to productivity scenarios, and cannot win just by diverting traffic from Douyin. Even if ByteDance wants to make efforts, there is no suitable target to bet on.
But personal agents are different, they are extremely to C, oriented to high-frequency and rigid-demand life services, and ByteDance and Doubao can finally go all out to make great achievements.
Different from AI programming and office Agents, personal agents are platforms that really need a large user base as support.
This is because programmers and enterprises can spend a lot of money on productivity AI, but ordinary people will not spend a lot of extra money on functions like AI ride-hailing; only when enough people use it can the high cost of personal agents be gradually diluted, and finally turn losses into profits.
The reason why Muse was first born in a super platform enterprise like Meta, rather than model giants such as OpenAI and Anthropic, is partly because Meta has a large enough user base. Douyin also has such conditions.
At the same time, life services are "user-agnostic", with strong universality, and do not require accurate user portraits. This also means that ByteDance's familiar traffic irrigation strategy has a clear place to play in the AI era.
Douyin and Doubao, the two core key players of the ByteDance ecosystem, both need to firmly seize the opportunity of personal agents.
As a giant in the mobile internet era, Douyin has always faced questions such as "how to transition to the AI era". Personal agents provide a solution: to act as the user source and service receiver for "capable AI that handles affairs".
In this way, Douyin's user value has a new conversion path and calculation method. Douyin does not need to do any extra work, and can bring hundreds of millions of daily active users into the AI ecosystem, and also activate its own mini-program ecosystem by the way.
Doubao's problem is how to stop being just a Chatbot. It has made many attempts before, including accessing AI e-commerce and other services; but from the perspective of logical smoothness and experience fluency, none of them can compare with the personal agent that it has just begun to try.
This personal agent battle is the most suitable for Doubao to fight, and it has the best chance to win. Under the circumstance that Chatbot is gradually becoming a thing of the past, transforming into a personal agent is the most urgent task for Doubao at the moment.
3
Doubao is making efforts on personal agents, with the entry point selected in travel, local life, and utility bill payment; the first to bear the brunt is likely to be Alipay.
As a high-frequency service, travel has always been a field valued by Alipay. In its AI life partner "Abao" at the lower left corner of the homepage, it has a separate "Travel" entry that aggregates multiple services such as online car-hailing, air and train travel, and public transport including buses and subways.
As for utility bill payment, it is even more Alipay's signature capability. It has spent a lot of time and energy polishing this type of capability, and its experience and penetration rate far exceed other similar services.
It can be said that these types of services are all core territories that Alipay cannot afford to lose.
In the past, WeChat, Douyin and other platforms have tried to attack, but they were easily repelled by Alipay. Now, Doubao, which has a smaller user base, is attacking again, and its long-term threat is even higher.
Objectively speaking, it is difficult for Doubao to shake Alipay's position in the field of life services in the short term. Alipay not only has more users, but its mini-program ecosystem is also far better than that of Douyin behind Doubao, and it is also supported by strong user mindset.
But looking one or two years ahead, with the popularization and prevalence of personal agents, user habits are likely to be gradually rewritten. At that time, Doubao's erosion of Alipay will officially begin.
The underlying logic is that the new generation of young people are AI-native, and they are more willing to "ask AI when in doubt", even "ask Doubao when in doubt", rather than rummaging through function options in traditional Apps.
Once Doubao completes the service closed loop, young people who trust Doubao deeply are likely to hand over their services to Doubao to complete together. It will not subvert a certain mature product immediately, but will let users gradually form new habits and forget the previous usage paths.
This subtle change has already occurred in vertical fields such as search, programming, and office work, leading to the revaluation of the value of corresponding enterprises. Alipay, which focuses on payment and life services, has no direct physical perception of this; now, with the arrival of the personal agent era, Doubao is about to attack the life service sector, and Alipay will have to stand on the front line of the battlefield.
Early users have shown unprecedented enthusiasm for personal agents. Although this enthusiasm has a component of novelty-seeking, it also reveals people's acceptance of AI "taking over" daily life.
Another variable is: according to Tencent's previous plan, WeChat Agent will accelerate its implementation and launch in the second half of the year. Compared with Doubao, WeChat's personal agent has an order of magnitude more users than Doubao, much higher user stickiness, and its WeChat mini-program ecosystem is by no means comparable to that of Douyin.
It can be predicted that when WeChat Agent is fully rolled out, local personal agents will usher in a real big explosion. After Doubao joined the battle, ByteDance has begun to show its posture of seizing the crown of personal agents; long-standing mature Apps are facing new major challenges.