Core Insights - The core objective of Yunzhisheng's AI medical layout is to promote equity in healthcare, addressing issues such as difficulty in accessing medical services, high costs, and uneven distribution of medical resources [1][2]. Group 1: Company Vision and Goals - The company aims to leverage AI medical capabilities to serve under-resourced areas, particularly in central and western regions of China, and to provide convenience for elderly individuals at home [2]. - The founder emphasizes the importance of creating "human-centered" AI that serves humanity and adds value, rather than replacing or controlling humans [2]. Group 2: Industry Challenges - The conversation highlights prevalent challenges in the healthcare industry, including the difficulty of obtaining quality medical care locally, which often necessitates travel to major cities like Beijing and Shanghai for treatment [2].

自动播放 出品 | 凤凰网财经《发现新势力》 黄伟：人最核心的需求无非是快乐和健康，用户的健康搜索需求印证了健康需求的重要性，这个事我们 一定会做。但医疗是非常严肃的领域，我们十几年前做医疗就坚持做严肃医疗，即便它门槛高、短期商 业回报低。中美医疗资源分布不同，中国最优质的医疗资源和医学知识财富都集中在公立三甲医院，各 医院还有专科特色。所以云知声会先和这些顶尖三甲医院合作，了解医生需求、提升模型能力，等模型 能力达到非常高的水平后，我们才会负责任地为 C 端用户提供相关健康服务。 本文来源为凤凰网财经《发现新势力》对话黄伟，完整视频内容在凤凰新闻客户端上线。 近日，云知声 (09678.HK) 创始人兼首席执行官黄伟在与凤凰网财经《发现新势力》对话时，谈及能否 用 AI 医疗能力满足用户健康搜索需求的问题，他表示该需求是明确的，云知声未来一定会布局相关业 务，但因医疗的严肃性，会先与顶尖公立三甲医院合作打磨模型能力，达到高水准后再向 C 端用户提 供相关服务。 （以下内容节选自《发现新势力》对话黄伟，对话实录有删减） 凤凰网财经《发现新势力》：PC 时代大家有大量医疗疾病类的搜索需求，这是很大的商业需求，你 ...

自动播放 出品 | 凤凰网财经《发现新势力》 近日，云知声 (09678.HK) 创始人兼首席执行官黄伟在与凤凰网财经《发现新势力》对话时，谈及大模 型 2.0 时代的行业企业格局，他将相关公司分为三类，指出不同类型企业依托自身资源能力呈现不同发 展态势，能否形成商业闭环成为关键发展因素。 （以下内容节选自《发现新势力》对话黄伟，对话实录有删减） 凤凰网财经《发现新势力》：在大模型 2.0 时代，你认为竞争格局是怎样的？ 黄伟：大模型 2.0 时代的公司大体能分为三类，第一类是 BAT 这样的巨头，手握算力、数据、场景、 人才等资源，做出大模型毫无悬念，还能实现模型与应用场景的闭环；第二类是像云知声这样的企业， 不做平台类业务，深耕特定应用场景，凭借够用的算力、领先的算法、多年行业积累和行业数据，也能 形成自身商业闭环；第三类是有技术和模型能力，但暂未找到真实应用场景的企业，这类企业若能找到 场景并形成模型与数据的商业闭环就能发展，反之可能消亡或被并购。 本文来源为凤凰网财经《发现新势力》对话黄伟，完整视频内容在凤凰新闻客户端上线 ...

在产品化层面，依托"山海·Atlas"智算一体基座，云知声将底层模型能力封装为标准化的智能体（Agent）解决方案，覆盖智慧客服、文档审核等高频场景， 显著降低了交付成本与实施周期。 与追逐泛娱乐应用的公司不同，云知声选择深耕医疗、医保、交通等对准确性、合规性要求极高的"严肃场景"。目前，"山海·知医"已在近400家医院部署， 700余家医院进入测试阶段。在北京友谊医院顺义院区，AI病历生成引用率接近90%，高复购率与扩展采购意愿验证了其不可替代的商业价值。 在AI大模型行业普遍面临商业化拷问的当下，云知声（09678）交出了一份罕见的高分答卷。公司最新公告显示，2025年其大模型相关业务收入预计达到6.0 亿元至6.2亿元，同比激增1057%至1095%，占整体收入比重提升至48%至53%。这一数据标志着大模型业务已从技术"试验田"彻底转型为公司的核心"现金 牛"，也预示着中国AI产业正从参数竞赛全面转向价值兑现的深水区。 未来，云知声计划将在医疗领域验证成功的"行业大模型+智能体平台+场景解决方案"模式，系统性复制到医保、政务、交通等高复杂度领域。通过智能体 的模块化组合，实现从"单点项目"到"平台级产品 ...

（以下内容节选自《发现新势力》对话黄伟，对话实录有删减） 凤凰网财经《发现新势力》：你觉得智能体在生产力方面的改造，会是一个什么状态？ 黄伟：智能体对生产力的改造是一种质变，它可以被看作是数字员工，拥有无数分身，效率和能力都优 于普通员工。对于中国这个产业大国来说，除了消费端的智能体，若行业智能体能得到大力发展，会进 一步增强中国制造业的综合竞争力，还能在部分行业缓解用工难题，同时解决当下的一些老龄化相关问 题。 出品 | 凤凰网财经《发现新势力》 近日，云知声 (09678.HK) 创始人兼首席执行官黄伟在与凤凰网财经《发现新势力》对话时，谈及智能 体对生产力的改造价值，他表示智能体对生产力的改造是一种质变，其作为数字员工具备高效、多分身 的特性，行业智能体的发展还将增强中国制造业竞争力，同时缓解用工和老龄化问题。 自动播放 ...

Core Viewpoint - The future competition in the AI industry over the next three years will center around intelligent agents, with a significant focus on multimodal development of large models this year, leading to rapid growth and widespread implementation of intelligent agents [1] Industry Insights - The intelligent agent sector is expected to experience rapid development and simultaneous growth in both consumer (C-end) and industry (B-end) applications this year [1] - The foundational capabilities of large models have reached a usable boundary over the past two to three years, resulting in a well-cultivated market and clearer demand for AI capabilities [1]

财经频道更多独家策划、专家专栏，免费查阅>> 责任编辑：山上 近日，云知声 (09678.HK) 创始人兼首席执行官黄伟在与凤凰网财经《发现新势力》对话时，谈及 AI 时 代文理科的专业差异与优势问题，他表示当下 AI 研发端理科生更具优势，但未来当 AI 能替代编程 后，审美、鉴别、人文等文科能力会变得更为重要。 凤凰网财经《发现新势力》：在 AI 时代，文科理科的专业差异是不是没那么大了，还是理科生更有优 势？ 黄伟：从 AI 模型研发的角度来看，理科生会更有优势。但从结果层面来讲，如果未来编程变得不那么 重要，AI 也能完成编程工作，那么文科就会变得很重要，因为后续发展所需的审美、鉴别、人文等能 力，都属于文科的能力范畴。所以当下从技术本身看，理工科更重要，而从未来的发展结果来看，文科 可能会更重要。 ...

Core Viewpoint - The discussion highlights the ongoing debate in the AI medical field, emphasizing that while AI has potential, it is not yet fully reliable, and the importance of continuing to train and value human doctors remains critical [1][2]. Group 1: Perspectives on AI in Healthcare - Huang Wei acknowledges that both Zhang Wenhong and Wang Xiaochuan present valid points regarding the role of AI in medicine, with Zhang focusing on the current limitations of AI and the need for ongoing doctor training [2]. - The core argument from Zhang Wenhong is that AI should not be fully relied upon at this stage, and there should be a pathway for doctors to grow and evolve in their roles [2]. - Wang Xiaochuan's perspective emphasizes patient autonomy and the ideal scenario where AI can assist without compromising the growth of medical professionals [2]. Group 2: AI Reliability and Patient Interaction - Huang Wei points out that many patients may use AI tools like DeepSeek to self-diagnose, but they often overlook the high error rates associated with such AI models, which do not take responsibility for their conclusions [2]. - The discussion illustrates the tension between the ideal use of AI in healthcare and the current reality where AI cannot replace the need for human doctors [2].

Core Viewpoint - The recent stock price volatility of Cloud Wisdom (09678.HK) is driven by high growth expectations in its large model business, technological breakthroughs, and commercial progress, despite ongoing losses and cash flow pressures [1][5]. Stock Performance - Cloud Wisdom's stock experienced significant fluctuations in February 2026, with a single-day increase of 16.77% on February 12, closing at HKD 345.40, followed by a 1.45% rise on February 13, closing at HKD 350.40 [1]. - Over the last five trading days, the cumulative increase was 20.33%, and over the last twenty trading days, it was 25.05%, although the stock is down 13.35% year-to-date [1]. - The stock exhibited a daily volatility of 11.58% on February 12, with a range volatility of 39.33% [1]. Business Progress - Cloud Wisdom has made significant advancements in the medical AI sector, with its "Shan Hai. Zhi Yi" large model achieving top rankings in three categories during the MedBench 4.0 evaluation [3]. - On February 6, 2026, the company entered a strategic partnership with Jia Hui Medical to expand into the high-end medical market [3]. - The company released the "Shan Hai. Zhi Yin Model 2.0," which supports full-duplex interaction and has achieved an accuracy rate of over 90% in recognizing complex background sounds [3]. Financial Situation - On February 13, 2026, the net outflow of funds was HKD 466.42 million from institutional investors and HKD 337.91 million from retail investors [4]. - The current price-to-earnings ratio (TTM) is -47.17, and the price-to-book ratio is 17.59, indicating a high market valuation expectation [4]. - Zhongsheng Group has set a target price of HKD 750.58 based on a 25x price-to-sales ratio for 2026 [4]. Future Development - The company faces risks related to slower-than-expected R&D progress and insufficient cash flow, with cash reserves projected to be only HKD 1.56 billion by the end of 2024, against a net loss of HKD 4.54 billion for that year [5]. - The customer retention rate in the medical business has declined from 70.4% in 2022 to 53.3% in 2024, indicating a need for improved customer stability [5].

Core Viewpoint - The discussion highlights the evolving nature of data privacy in the context of AI development, emphasizing the potential benefits and risks associated with the creation of personal digital assets and the transparency it brings to individuals' interactions with AI [1][2]. Group 1: Data Privacy and AI - In the internet era, data sharing has been prevalent, but the scope of data being shared has expanded significantly, leading to concerns about the loss of boundaries regarding personal data [2]. - The concept of transforming all personal data into a digital asset is seen as a positive development, provided it is done securely, allowing individuals to become "super individuals" with AI that understands their needs [2]. - However, there is a contrasting risk where individuals may become completely transparent to AI, which could predict their actions and decisions, raising ethical concerns [2].