Economic Observer reporter Ren Xiaoning
Pan Yuefei recently noticed that TodayAI (hereinafter referred to as “Today”) “has some skills.” Previously, he took on a commercial project worth 100,000 yuan, but by the agreed payment date, the other party failed to make the payment. Last month, they called him and mentioned that there were issues with the original contract, and a new process needed to be followed, so they agreed on a new payment date. On October 20th, Today reviewed his recording and realized that the payment deadline had passed, so they proactively reminded him to follow up on this outstanding debt.
Pan Yuefei is the person in charge of three small and medium-sized companies and is on business trips on a weekly basis, with his time being very fragmented. After using Today, he believes that this product can basically meet his needs for a personal AI (artificial intelligence) assistant.
Personal Agent has been one of the hottest tracks in the AI field since September this year. Muse launched 12 days ago and achieved 640,000 daily active users on the mobile platform in the United States, nearly three times the number at the initial stage of launch for ChatGPT mobile applications. Instinct has a valuation of over ten billion dollars just one year after its establishment, and within three weeks of launching, 40% of its users voluntarily granted credit card usage permissions.
Domestic internet giants such as Alibaba, Tencent, and ByteDance have already launched or are about to launch similar products. Alibaba's Qianwen GO was launched on October 13th, with advanced features like "My Applications" and "Cloud Pro" currently in the beta testing phase. Tencent's personal Agent, codenamed HandyBot, has quietly launched a service account on WeChat, labeled as " YourPersonalAIAgent", with Shenzhen Tencent Computer System Co., Ltd. as the operating entity. There are plans to launch an independent App in the future. The ByteDance DouBao team is also advancing related projects, with the product named Xiaodou.
An insider from Qianwen told reporters that although Qianwen has not been online for a long time, it is a strategic business for Alibaba. Within Qianwen, it is displayed alongside the “quick mode” for daily conversations.
Many people believe that personal Agent has the potential to become the third commercially viable scenario in the AI arena, following AI video and code generation (coding). Even more optimistic entrepreneurs and investors hope it will grow into the "WeChat of the AI era."
However, at this stage, the number of users for personal Agent products is not large. The more well-known Muse has accumulated 10 million users in one and a half months. Domestic products are still mainly spreading within a small circle in AI and among geek communities.
Trial personal Agent
Pan Yuefei first used Today on October 12th. He is also a beta tester for several other personal Agent products, including Lingguang, Grok, Dots, etc. He discovered Today through his own search and learned that it was a product created by Qi Junyuan, the former person in charge of Doubao PC. Therefore, he decided to try out this product.
Pan Yue has granted a large number of personal permissions to Today, covering more than twenty tools that he uses frequently on a daily basis, including calendars, email, DingTalk, Lark, WeChat Reading, Kingsoft Document, Tencent Meeting, and others.
In addition to that, he also installed a third-party Skill on his own and connected it to the Plaud AI recording product. From this, it is possible to know who he met every day and what topics were discussed.
One night at 11 o'clock, when he returned home from a business trip, Today took the initiative to ask him if he had just come back from Shanghai, and also reminded him that his wife's birthday was approaching.
Calendar is the feature that he uses Today the most. Today reads his schedule information, helps him organize and refine his travel plans, and reminds him of flight and high-speed train times. In the past, he had to sit in front of a computer, flip through travel apps, and manually enter and cross-check the information. Now that these tasks are handled by AI, he estimates that he can save about 20 minutes per day.
After downloading Today, he uses it three to four times a day on average.
Today comes from a startup company, developed by Huihe (Shanghai) Intelligent Technology Co., Ltd., which was established in December 2025 with Qi Junyuan as the legal representative and a registered capital of 1 million yuan. Qi Junyuan also owns thisjian Wuxian (Shanghai) Intelligent Technology Co., Ltd., with a registered capital of 10 million US dollars and established in March 2026.
On October 21st, Wuyuan Capital announced that the Today team has completed over $200 million in financing in total. The investors include IDG Capital, Tencent, Wuyuan Capital, Ant Group, Sequoia China, Transsion Holdings, Tongge Venture Capital, and Shanghai Jiao Tong University Hanyuan Assets.
The reporter made several calls to Today to request an interview, but was refused on the grounds that they needed to contact the person in charge of the marketing department.
Today founder Qi Junyuan once served as the person in charge of ByteDoubao PC. In his early years, he founded Teambi – tion (which was later acquired by Alibaba), and also held positions such as head of Alibaba Cloud Disk and vice president of Lark products. Within two days of its establishment, Infinite Company officially announced the completion of a $10 million angel round of financing. IDG participated in the investment, with IDG providing the underlying models.
TodayAI started its internal testing in July this year and officially launched on September 24th. The number of registered users has not been made public yet.
Apart from Today, there are already a number of domestic startups on the market that have launched personal Agent products, including ManusCue, Magic Space Mooli, LinkAI / CowAgent, WishSeen, etc. There are even more personal A-gent products overseas; the most well-known one is valued at over ten billion US dollars. In addition, there are Lindy, Or – bits, as well as some specialized tools for specific scenarios: FerryHealth for helping clinics schedule consultations, Miso for travel planning, and Stanley for content creators.
BAI Senior Capital Partner Wang Tianfan has invested in three companies that are focusing on personal Agent. These companies started their R&D efforts half a year ago, and their products were first launched in overseas markets. Among them, RabbitOS3 was released on September 22 this year, and on the 7th and 14th, it achieved a 60% user retention rate, with a rapid increase in the number of users. Wang Tianfan stated that the user base of this product is on par with that of domestic competitors such as Today, with each company having its own focus and approach.
Wang Tianfan has long been optimistic about the personal Agent sector. In his view, such products are capable of continuously managing user context and producing results that better meet user needs. More importantly, when the marginal cost of Token drops to a sufficiently low level, personal Agent will unlock a vast array of personalized demands. By then, this will become a much larger market than AI videos and coding; it could represent a business on par with WeChat's scale.
Today's competitors are not just startups, but also large corporations.
At the Alibaba Cloud Qiyi Conference in September this year, Zheng Sishou, the product manager of Qianwen Business Unit, stated that Alibaba's goal is to turn Qianwen into "your personal Agent".
Zheng Sishou said that in the end, personal Agent is not just about passively executing tasks and catering to users, but rather gradually developing a capability to understand users' long-term goals, comprehend their inner personalities, identify the discrepancies between what users verbally express and their true interests, and provide reminders at critical points while negotiating with them.
More suitable for large companies
The individual Agent is currently very popular in the AI circle, but when reporters asked several domestic investors who focus on the AI field, most of them did not place their bets in this area.
Zhang Yu, a founding partner of Qingzhi Capital, said that he has recently reviewed several startup projects involved in the upstream and downstream of personal Agent. His direct impression is that this is not an easy task to undertake, as it requires a huge amount of capital investment. Another investor believes that platform-level opportunities in the AI application layer are most likely to be seized by large companies.
An investor from a leading domestic investment institution, AI, began to look at personal Agent projects last year. At that time, a number of startup teams on the market were focusing on desktop versions of personal Agent; most of them were in the internal testing or gray-scale phase, with user bases in the hundreds of thousands range.
Looking back, these early projects are not going smoothly at the moment. There are several reasons for this: there is less user demand, the target audience is narrow; the interaction threshold is high, and technology has not kept up; and users are not very willing to pay. He believes that those who can truly create platform-level products must have at least experience as the number one in product or technology at a large company.
After using Today, Pan Yuefei concluded that the most likely outcome for this product would be to be acquired by a large company. Comparing it with several similar products, he believes that Today does not have any disruptive breakthroughs in terms of product or technology, nor are there any insurmountable technical barriers.
Although Today has already helped him solve some of his problems, there are still many needs that cannot be met. For example, he hopes that his personal Agent product can integrate with high-frequency App platforms such as WeChat, Taobao, Meituan, Kuaike Cloud Disk, and QQ browsers, but currently, Today cannot achieve this.
At this stage, the products that are more closely connected are mainly Todoist and Notion which are overseas products. The ecological coverage of App with hundreds of millions of domestic traffic is significantly insufficient.
Pan Yuefei most hopes that one of his personal Agent products can be integrated with WeChat, as 70% of his high-value private data is stored on WeChat. However, WeChat is cautious about such AI automated operation tools. When DouBao Mobile was first launched, WeChat was the first to reject its integration; subsequently, Taobao and Xianyu also displayed human-machine verification to block such calls, while Agricultural Bank of China and China Construction Bank directly terminated related services.
When the second-generation DouBao phone was released this year, more than 150 APP refused to allow it to perform any operations. "Individuals Agent cannot do without an ecosystem, and an ecosystem either comes from hardware or relies on a mature product matrix; large companies are more capable in this regard," said an investor from a leading domestic investment firm mentioned above, AI.
Muse's success in the United States is closely related to the ecosystem of Meta. According to Apptopia data, over 95% of Muse users also use Facebook, and 63% use Instagram, indicating a high degree of user overlap.
Among the products that have already been launched in China, there is Qianwen GO! It focuses on meeting the needs of daily life, work, and study scenarios, including providing services such as price comparison for discounts and reservation of restaurant venues; editing and generating Office documents, processing data, and other work-related services; as well as offering professional academic paper tools and other learning functions.
Among them, "cloud Pro" supports accessing relevant information from Taobao, Damai, Gaode, and Kuak Cloud Disk after user authorization. The aforementioned Alibaba insider stated that currently, Qianwen GO is mainly connected to Alibaba's own ecosystem. To integrate with more ecosystems in the future, it will be necessary to negotiate with other companies one by one.
Qianwen GO! Two user cases that received a lot of attention are as follows: One user couldn't remember which flight was delayed, so Qianwen GO! verified 4 orders from 4 airlines and identified China Eastern Airlines MU2445 as the responsible party. When the customer service offered a voucher as compensation, the user clearly replied, "I firmly refuse the voucher and only accept cash compensation," and ultimately received 70 yuan in cash, which was transferred to their Alipay account; another user didn't know how to apply for Taobao's price protection, so Qianwen GO! found the price protection portal for them, searched through their historical orders one by one, and helped the user recover the 52 yuan difference resulting from the discounted non-stick pan.
Wang Tianfan believes that the competitive path for personal Agent in China will be different from that overseas. Major companies such as Tencent, Alibaba, Baidu, and REDnote are likely to launch their own personal Agent products. However, the products developed independently by these large companies may not necessarily be what users truly desire. He thinks that DouBao, which has the largest user base, could very well become the biggest winner in this domestic market if it enhances its context management and calendar scheduling capabilities.
WeChat-level business
The aforementioned leading investor from AI remains optimistic about this sector: "Individuals are the hub of all information; they are equivalent to WeChat in the AI era. It's a big business, and we are definitely bullish." However, who will actually be able to bring this product to market remains an unknown.
The difference between this round of personal Agent and previous products lies in the significant improvement of AI's capabilities, which now include long-term memory, allowing it to help users handle more daily trivial tasks; many products are now adapted for mobile devices, making them more convenient than the desktop versions of the past.
The number of overseas Muse users is still growing. According to data estimated by Similarweb, one month after its launch, Muse had approximately 2.18 million daily active users on October 5th. The founder of Instinct mentioned that payments handled through Instinct are expected to exceed $1 billion within a year. Instinct indicates that 50% of consumption authorization requests are for travel scenarios, and within three weeks, 40% of users voluntarily submitted their credit card usage permissions.
At present, most individual Agent products have not been fully launched: Muse is only available in regions such as the United States and Canada; Instinct adopts an invitation system; Qianwen GO is still in grayscale testing, with users waiting in line for reservations.
A report by Morgan Stanley estimates that the monthly service cost per Muse user ranges from $3 to $130, with a baseline scenario of $37, which translates to an annual cost of $445. This is already higher than the average annual revenue of about $70 for Meta global users. The report also mentions that Muse will not truly achieve scale commercialization until at least the end of 2027 to 2028.
"This is a very expensive market segment; currently, the monthly service cost for a single user ranges from $10 to $100. It's quite challenging for startups to sustain a large-scale free model," speculated Wang Tianfan. He suggested that personal Agent products are likely to take a high-end approach, targeting users with a strong willingness to pay and offering more advanced context processing capabilities.
The aforementioned investor with the ID AI also believes that there is currently no product that can reduce costs while ensuring a good user experience. Whether it's for enterprises or individuals, the cost of computing power is a significant expense. He predicts that it will not be until the fees for high-quality Token services are reduced to a level comparable to that of 4G data usage, and people no longer find paying for Token services so painful, that the personal Agent market will experience a surge, and only then will there be a possibility of a AI version of WeChat emerging.
(Author: Ren Xiaoning)
