Ant Group's foray into AI healthcare: Taking the narrow path for a long - haul journey?
The "inclusive" development of technology is not a romantic tale that can be achieved overnight, but rather a systematic project composed of countless "small things".
< p > Another big player has started to sprint full - force in the field of medical AI. p >
< p > On June 26th, after 11 years in the medical field, Ant Group officially launched the independent AI health application "AQ". This is a brand - new upgrade of the "AI Health Butler" that was launched within Alipay last September. In terms of functionality, it not only offers regular services such as medical consultations and health science popularization, but also directly connects with thousands of domestic hospitals, millions of doctors, and hardware device manufacturers like Yuwell and Huawei, forming a comprehensive health management system covering both in - hospital and out - of - hospital periods. p >
< p > In other words, in the future, if you want to ask an AI about health issues, handle cross - regional medical insurance reimbursement, or get advice on blood sugar and blood pressure control, you no longer need to switch back and forth between different AI assistants and apps. One AQ can handle it all. p >
< p > Zhang Junjie, the vice - president of Ant Group and the head of the Medical and Health Division, said, "AQ will be Ant's new gateway to the development of AI - powered medical and health services. In the future, Ant will also collaborate with more doctors, hospitals, and medical - related enterprises to provide more inclusive medical and health services to users through AQ." p >
< p > It's worth noting that this is not Ant's only recent move in the medical field. In the annual ESG report released by Ant Group on June 30th, "Inclusive Medical and Health Services" was upgraded to one of the group's five major business segments for the first time. p >
< p > What might these changes imply? Why is Ant betting on the AQ application and accelerating its exploration of AI - powered medicine? p >
< p class="image-wrapper">
p >
< h2 >< strong > 01. Why Does AQ Exist? strong > h2 >
< p > After the release of AQ, many people asked Zhang Junjie a question: Since Alipay is already a powerful traffic entry point, why does Ant want to develop a new, independently - operated app? He believes this is closely related to "what role Ant plays in users' health needs". p >
< p > Initially, this role was a "convenient payment tool". 11 years ago, Alipay pioneered the first online registration and payment service in the country. After that, it carried out payment entry reforms for one hospital after another. As more people registered for appointments on their mobile phones, fewer people queued overnight in hospitals. In 2019, Alipay launched the medical insurance code. Now, using medical insurance for medical treatment has become a daily routine for nearly 1.2 billion people, 700 million of whom are Alipay users. Later, this role became a "medical and pharmaceutical assistant". As diverse user needs such as medical consultations, medicine purchases, and insurance reimbursements emerged, "linking" became the core. Thus, from a single medical insurance code, the Alipay Medical and Health Channel was developed to carry more functions. p >
< p > "In the past ten years, we spent time solving small problems for the public in medical treatment, such as reducing queues and simplifying procedures. However, these only made things more convenient and did not really address the difficult problem of making medical resources more inclusive," Zhang Junjie said. p >
< p > Now, with the arrival of the AI era, this technological path that directly points to productivity transformation is most likely to solve the long - standing problem of medical inclusiveness. So, Ant began to re - think its role and mission. p >
< p > Since the beginning of this year, major medical institutions have actively integrated with DeepSeek, and top doctors have also started using AI tools. The medical system is embracing AI more comprehensively than ever before. It's no exaggeration to describe the professional AI applications and services for medical institutions as "springing up like mushrooms after rain". p >
< p > However, on the other hand, ordinary users have great needs but rarely feel the changes in efficiency and experience brought by AI. The proportion of sub - healthy people in China exceeds 75%, and the potential population for chronic disease management is estimated to be 500 - 600 million. Referring to the development trend of developed countries, this figure will gradually increase as the population ages. Compared with the growing demand for health management, there are too few AI health tools for ordinary users. p >
< p > For Ant, if it can integrate different forces from upstream and downstream, it can theoretically bring about greater changes. "The 'end - game' we envision is that everyone has a trustworthy health butler. In the past, the concept of AI was unclear, and our concept of the butler's image was vague. Now we think it should be like a service team that can handle everything for users," Zhang Junjie mentioned. p >
< p > In September last year, such an AI health butler took shape within Alipay. It included AI health services such as finding doctors and reading reports and connected nearly 200 celebrity doctor AI avatars to build a service ecosystem. Without significant marketing promotion, it served over 70 million users in 10 months. Besides real positive feedback, users also continuously reported issues like "the entry is too deep and hard to find" and "why can't I see the push messages from the health butler". p >
< p > So, the team realized that a "health butler" hidden among Alipay's numerous functions was no longer sufficient, and an independent app became a necessity. "This was a natural decision," they said. p >
< p > For Ant, this actually reflects a transformation from "being passively searched by users" to "actively providing services to users". "A health butler needs to be able to actively reach out to users, establish a connection with them, and tell them that health is important and should not be neglected. From this perspective, an independent app will provide a better user experience." p >
< p > In terms of actual effects, compared with the previous AI health butler embedded in Alipay that required users to search for it, AQ first brings a different interactive experience: AQ has a simpler interaction form and interface, and can integrate all services through just a conversation. The app also lists several core functions such as AI medical consultations, report interpretations, doctor recommendations, and medical insurance Q&A on the home - page dialog box. At the top of the page, users can switch to another service interface where nearly 200 doctor AI avatars and multiple health intelligent agents "wait" to provide 24/7 services. p >
< p class="image-wrapper">
p >
< p class="img-desc">Ant Group's AI health application AQ p >
< p > In terms of specific services, AQ has also been upgraded. For example, in the past, it only provided simple answers without follow - up questions. Now, it has been updated to the "clinic mode". The system will lead users to the AI clinic based on their questions, actively conduct multiple - round follow - up questions like a doctor in a consultation, and provide an analysis of possible conditions and suggestions, which is very similar to the experience of seeing a doctor in a hospital. p >
< p > Since AQ aims to provide consultations and suggestions and cannot replace the diagnosis and treatment of medical professionals, after an AI medical consultation, the system will also actively promote functions such as online doctor consultations and appointment registrations, guide users in need to enter the offline medical treatment process, and provide continuous follow - up services such as cloud - accompanied consultations, medical insurance Q&A, and health management. p >
< p > Another change is that AQ is connected with hardware devices from Yuwell, Sanuo, Silicone Base Dynamic, Huawei, Apple, etc. After users actively authorize, it can intelligently analyze users' health data such as blood sugar, blood pressure, breathing, and sleep through wearable devices and provide comprehensive and considerate suggestions to help users manage their health. p >
< p > AQ is Ant's attempt to get closer to the user end in AI - powered medicine. "Putting users' interests first" was the principle set at the beginning of the project. Zhang Junjie said, "Therefore, we won't be overly concerned about the number of users of the independent AQ app for now. At this stage, we just want to offer a more comprehensive product for people to choose from, so that users with different needs can enjoy suitable and the best service experiences. This is our only goal." p >
< h2 >< strong > 02. Why Do Doctors and Hospitals Choose AQ? strong > h2 >
< p > Different from Ant's previous medical products, the direct audience of AQ is C - end users. However, it's a common understanding that doing C - end business in the medical industry is not easy. Why does Ant dare to take this step? What are AQ's core advantages? p >
< p > First, in terms of technology, AQ still relies on the technology base centered on Ant's medical large - scale model. According to reports, this model has certain leading advantages in three dimensions: the construction of a medical multi - modal data system, the innovation of model architecture design, and real - world scenario applications. p >
< p > For example, in terms of data, it covers trillions of tokens of professional medical text corpora, tens of millions of knowledge graphs, and multi - modal data, and has a deep understanding of disease characteristics, diagnosis and treatment logic, and medical expressions. In scenarios, Ant's medical large - scale model has an accuracy rate of over 90% in recognizing images of reports, drugs, and skin diseases in the AQ product application, and can recognize and interpret more than 100 types of complex, multi - page medical test reports. p >
< p > As evidence of the model's capabilities, in the HealthBench evaluation led by OpenAI, Ant's medical large - scale model outperformed mainstream general large - scale models such as DeepSeek in two core indicators: complex reasoning ability (HealthBench Hard score of 0.2702) and medical consensus ability (HealthBench Consensus score of 0.9275). In the list updated on June 24th by the domestic authoritative medical large - scale model evaluation platform MedBench, Ant's medical large - scale model ranked first again with a score of 98.7. p >
< p > Second, in realizing medical capabilities, AQ not only benefits from the mutual support with doctor platforms like Haodf within Ant but also relies on the in - depth co - creation and participation of thousands of medical professionals inside and outside the company. Leaders from different disciplines of more than 10 top - tier tertiary hospitals have also formed a professional advisory group and deeply participated in work such as the evaluation of AQ's specialty Q&A to ensure the professionalism and rigor of services such as health consultations and medical science popularization. p >
< p > Chi Chenfei, an attending physician in the Department of Urology at Shanghai Renji Hospital, participated in the development of AQ's urology specialty model and intelligent agent. He recalled that when the hospital decided to embrace AI, it knew "this would be a very challenging task" because training the model not only required materials such as medical guidelines and pathological data but also needed to learn how top doctors "make decision - making judgments based on clinical experience, and each doctor is different". To "extract" these experiences, it took "several months" just to build the corpus. Then, they designed various virtual patients for doctors to conduct consultations and "taught the model step - by - step" based on the Q&A process. p >
< p > Currently, the urology specialty intelligent agent "RJUA" has been operating within AQ for nearly a year. "I think specialty intelligent agents are the most valuable among all medical intelligent agents. Because as long as it is deep enough, it can effectively solve some medical problems. Evolving from specialty intelligent agents to individual doctor intelligent agents is also like the process of climbing from being a medical student to being a doctor," Chi Chenfei said. p >
< p > The doctor intelligent agent is one of the highlights of AQ. When the AI health butler was launched on Alipay last year, there were only 2 doctor intelligent agents. Now, this number has increased exponentially. Nearly 200 doctors, including two academicians of the Chinese Academy of Engineering, Wang Jun and Liao Wanqing, and those from Peking Union Medical College Hospital and the Second Affiliated Hospital of Zhejiang University School of Medicine, have launched their intelligent agents on AQ to answer users' questions 24/7. p >
< p > Doctors who were difficult to make an appointment with even through "scalpers" in the past are now gathered in the small AQ app, benefiting more users. Mao Hongjing, the deputy dean of the Mental Health Center Affiliated to Zhejiang University School of Medicine, mentioned that in the past, he could only receive 600 patients a month at most. Now, through the "AI avatar", he can serve over 110,000 people a day, and the service scope has expanded from within the province to the whole country. "If we can make it possible for places like Tibet, where medical technology is relatively weak, to access better medical resources, this is our original intention." p >
< p > Qin Wenxing, a chief physician at the Cancer Hospital Affiliated to Fudan University, believes that AI applications actually expand the possibilities of medicine and create medical paths that were unimaginable in the past. His "AI avatar" on AQ once received a late - night call from a patient: A patient with advanced breast cancer poured out her treatment difficulties to the "doctor avatar" at night. The system immediately launched a crisis intervention process. By exploring her psychological support point of "wanting to see her daughter graduate from college" and presenting prognosis data visually, it finally helped the patient rebuild her confidence in treatment. "This kind of anytime, anywhere psychological support is also difficult to achieve in traditional medicine," Dr. Qin said. p >
< p > "Respect professionalism and work steadily" is the warning Zhang Junjie gives to himself and his team. This may also explain the key to making the AQ product successful: Although technical strength is important, professional refinement requires cooperation with doctors and medical institutions. p >
< h2 >< strong > 03. How Far Can the AQ - style "Open Platform" Go? strong > h2 >
< p > Since 2025, the market competition in medical AI has continued to intensify. There's no need to mention established big players. New - comer AI companies are also not willing to lag behind. However, from the current competitive landscape, technology - driven players with an AI background are often strong in algorithms but weak in scenarios, while traditional medical enterprises have channels but lack the ability to iterate. And platform giants also need to consider the balance of ecological cooperation. p >
< p > In other words, a single - point breakthrough can no longer meet the needs of this special medical track. A complete ecological chain covering "technology base + scenario implementation + regulatory compliance" must be built. p >
< p > If we look at it from an industry perspective beyond Ant, AQ actually presents a possibility for the future of medical AI: When the technology base is solid enough and the ecological connection is wide enough, can medical AI break through the limitations of being just a "tool" and truly become a "new platform" that renews the paradigm of medical and health services? When medical AI products evolve from "tools" to "partners", will the relationship between humans and AI be rewritten? p >
< p > It's true that AQ's open ecosystem has taken shape, but challenges still exist in the future. For example, in terms of the issue of responsibility boundaries, how should the responsibility be divided when AI gives inaccurate medical advice? Another example is how to maintain the sustainability of the business model. These questions may only become clearer through future practice. p >
< p > Zhang Junjie mentioned that if we evaluate projects based on a one - year time frame, most things shouldn't be done. For example, it took nearly 2 years to create specialty intelligent agents with hospitals and doctors, 5 years to standardize and scale medical insurance payments, and nearly 10 years to digitize services such as appointment registrations and consultations in national medical institutions. "But once these things are accomplished, hundreds of millions of people will benefit." p >
< p > In Ant Group's latest ESG report, in addition to digital payment and lifestyle services, digital inclusive finance, technology industrialization, and globalization, "Inclusive Medical and Health Services" has also been upgraded to one of Ant's five major business segments. p >
< p class="image-wrapper">
p >
< p class="img-desc">Ant Group's 2024 Sustainability Report p >
< p > If we look at Ant's 11 - year exploration in the medical and health field, from registration and payment, medical insurance codes to the AI health butler, it may prove that technological "inclusiveness" is not a romantic story that can be achieved overnight but a systematic project composed of countless "small things". p >
< p > For example, enabling