Core Insights - Qianwen has launched an AI app that can perform real-life tasks, marking a shift from conversational AI to practical applications in daily life [1][2] - The app integrates with Alibaba's ecosystem, allowing users to order food, shop online, and book travel services through simple voice commands [1][2] - The upgrade includes over 400 AI functionalities, indicating a significant advancement in AI capabilities to assist users in real-world scenarios [1][2] Group 1: AI Functionality and Integration - Qianwen's app can recommend popular food options and facilitate orders through integration with Taobao and Alipay, showcasing its ability to streamline the ordering process [2][3] - Users can receive tailored product recommendations based on their queries, leveraging Taobao's extensive database and review system [3] - The app can handle complex requests, such as planning a hiking trip by suggesting necessary gear and directly linking to purchase options [3] Group 2: Strategic Importance and Market Position - Alibaba views Qianwen as a key player in the AI-to-C (consumer) market, aiming to create a comprehensive AI lifestyle interface [5][6] - The app's launch is part of a broader strategy to integrate AI with Alibaba's various services, enhancing user experience and operational efficiency [5][6] - The company is positioning itself to capitalize on the growing demand for AI-driven consumer solutions, with expectations of significant market growth in the coming years [4][5] Group 3: Competitive Landscape - Other tech giants like Baidu and ByteDance are also developing similar AI capabilities, indicating a competitive race in the AI-to-C space [7] - The integration of AI with real-world applications is becoming a common strategy among leading internet companies, aiming to create a feedback loop for continuous model improvement [7]

Core Insights - China is at a critical stage for future industry development, with the "14th Five-Year Plan" emphasizing the exploration of diverse technological routes, typical application scenarios, and feasible business models to promote quantum technology, biomanufacturing, and embodied intelligence as new economic growth points [1] Group 1: Future Industry Development Strategies - Local leaders have significantly increased their focus on future industries, with over 700 relevant event information monitored, covering more than 1,700 companies [1] - The development strategies for future industries are becoming clearer, focusing on future manufacturing and future information as core drivers, integrating key technologies like large models and embodied intelligence [1] - Future materials, future energy, and future space are identified as important supports, accelerating breakthroughs in key technologies and practical applications [1] Group 2: Key Future Industries - Future manufacturing emphasizes a dual-drive approach of "technology sourcing + pilot testing" to solidify the foundation of new productive forces [4] - Future information aims to strengthen computing power and expand AI applications, creating a "AI + connectivity + computing power" ecosystem [4] - Future health focuses on the integration of medical and engineering fields, exploring new industrial paths that combine cutting-edge technologies with clinical applications [4] Group 3: Strategic Framework for Future Industries - The strategy highlights localized layouts to solidify the strategic foundation of resource-driven future industries [5] - It advocates for a gradual approach to implement long-term strategies for future industries [5] - The framework emphasizes the importance of innovation services to support disruptive technology sourcing and the creation of an innovative ecosystem [5]

Core Insights - The automotive industry is undergoing a profound transformation driven by artificial intelligence (AI), moving from early concept validation to practical implementation, reshaping the entire industrial ecosystem from technology development to business models [1][2]. Group 1: Restructuring the Automotive Value Chain - The core of the AI era is "cognition," contrasting with the "computation" focus of the information age, characterized by data-driven societal mechanisms, algorithm-influenced decision-making, and human-machine collaboration [4]. - AI systems possess learning, reasoning, and self-optimization capabilities, participating as "cognitive partners" in production, management, and service processes, significantly impacting manufacturing, transportation systems, and urban operations [4][5]. - Multiple countries have elevated AI to a national strategic level, promoting technological innovation and industrial application integration through systematic policies [4]. Group 2: Technological Pathways in Autonomous Driving - Three differentiated technological routes have emerged in the global autonomous driving sector: 1. The robust route represented by Waymo, utilizing a combination of lidar, radar, and cameras, achieving L4-level commercial operations but facing high costs and slow scalability [6]. 2. The aggressive route advocated by Tesla, relying on a "pure vision" approach with lower hardware costs, primarily applied in L2+/L3-level driving assistance, but facing challenges in extreme scenarios [6]. 3. The system redundancy route explored by Zoox, focusing on physical redundancy in vehicle structure and sensor layout, aiming for L4/L5-level fully autonomous driving but with a longer commercialization cycle [6]. - Autonomous driving technology has demonstrated high reliability in practical applications, such as Waymo vehicles operating smoothly in complex urban environments, enhancing passenger safety and trust through transparent real-time interaction [6][7]. Group 3: Future of Autonomous Driving - There is no "one correct" technological answer for autonomous driving; companies must balance safety, cost, scalability, and reliability [7]. - As AI agent capabilities evolve, computing costs decrease, and regulatory frameworks improve, autonomous driving is expected to expand from pilot projects to broader applications [7]. - Future competition will focus not only on sensors or algorithms but also on system engineering capabilities, data loop construction, and long-term technological patience, marking the beginning of a significant revolution in mobility [7].

Core Insights - The introduction of the AI model DeepSeek by Deep Exploration Company in early 2025 has sparked a significant application boom in the financial industry, marking a transformative technological force comparable to the mobile internet [1] - The banking sector is leading the procurement of large models, with a notable increase in project numbers and funding, indicating a shift in focus from computational power to application effectiveness [3][5] Group 1: Market Trends - In 2025, the financial industry saw a dramatic increase in large model procurement, with 587 projects awarded, representing a 341% year-on-year increase in project numbers and a 527% increase in disclosed funding to 1.506 billion yuan [3][5] - The banking sector accounted for nearly half of the total projects with 290 projects, and 75.2% of the total funding, establishing a dominant position in the market [5][6] Group 2: Project Distribution - The distribution of project types in the financial sector for large models in 2025 shows that banking projects comprised 49.4% of the total, with disclosed funding of 1.13221 billion yuan [6] - The focus is shifting from computational power projects to application projects, with application-type projects (including intelligent agents) rapidly increasing in number and becoming the primary procurement direction [7] Group 3: Driving Forces - Multiple factors are driving the banking sector's embrace of large models, including supportive government policies aimed at accelerating the intelligent transformation of the financial industry [8] - The maturity of technology has reached a turning point in 2025, with significant improvements in the accuracy, reliability, and practicality of large models, particularly with the rise of open-source models like DeepSeek [8][9] Group 4: Competitive Landscape - The competitive pressure in the banking sector, characterized by narrowing interest margins and intensified competition, necessitates new tools for efficiency and differentiation, with AI applications potentially reducing costs by up to 70% in certain categories [9] - Customer expectations for financial services are rising, demanding quicker responses and more personalized experiences, which traditional technologies struggle to meet [9] Group 5: Application Scenarios - Specific application scenarios in the financial sector are becoming concentrated, with intelligent customer service and digital personnel leading the number of awarded projects [10] - The focus on intelligent agents is increasing, with 49 projects explicitly mentioning "intelligent agents," indicating a growing interest in embedding AI capabilities into specific applications [11] Group 6: Future Outlook - As the application of large models deepens, the procurement of application-type projects is expected to grow, with banks likely to develop their own intelligent agents based on clear scenarios and engineering capabilities [11][12] - The financial industry is seen as a data and service-intensive sector, with significant potential for further exploration and application of large models [12]

专题：2025科技风云榜 专题：2025科技风云榜 "2025科技风云榜"年度盛典于2026年1月15日在北京举办，今年活动主题为"启新智，赴新程"。 欧洲科学院院士、西湖大学可信与通用人工智能学院创始人金耀初发表《探路AGI：走向类脑具身智 能》主题演讲，谈及如何让具身智能成为研究或者通向通用人工智能的途径，金耀初认为，需要更多结 合人脑工作的机制。 他表示，人脑有860个神经元，有更多的连接。但是它的功耗才20瓦左右，和现在大模型相比差得非常 大，它的能耗非常低；人的基因数量2万-2万5千个，再加上各种各样的复杂器官，为什么只有2万多个 基因就能够编码这么复杂的系统？从这些角度都值得我们探讨。 金耀初指出，人脑是世界上目前最为先进的智能系统，有很多工作的机制，和我们目前的大模型是不完 全一样的。目前大模型和人工智能的深度学习模型一般是有输入，中间做信息处理，最后输出，只是单 向的信息流动。而人脑至少是有自上而下，自下而上更多信息处理的通道，而且它的功能结构是经常在 发生变化。大模型预训练好之后可能做微调，但是不会出现不同问题的时候会采用不同的结构，人脑有 这个能力。通过基因调控，神经调控很多的机制，不断改 ...

Core Insights - Alibaba's Qianwen App has integrated with Taobao Flash Purchase and Alipay AI payment features, allowing users to place food orders through simple text commands, streamlining the ordering process [1] - The app's "Task Assistant" feature is being tested, enabling users to handle more complex ordering tasks and various other applications, enhancing user experience and efficiency [3] Group 1: Integration and Functionality - Qianwen App allows users to order food by simply typing commands like "help me order a cup of milk tea," utilizing Taobao Flash Purchase for precise location and merchant matching, and Alipay AI payment for seamless transactions [1] - The integration enables users to receive coupons when ordering through the Qianwen App, enhancing customer engagement and incentivizing usage [1] Group 2: Task Assistant Capabilities - The "Task Assistant" can manage complex orders with multiple conditions, such as "10 cups with ice, 10 cups without sugar," and is currently in a targeted testing phase [3] - It can efficiently handle repetitive tasks in various scenarios, such as generating annual expense reports from uploaded invoices, creating analysis reports from data screenshots, and producing wedding invitations without coding [3][5] Group 3: File Processing and Efficiency - Qianwen App can process up to 100 files on the web and 10 files on the app, significantly improving efficiency in financial, legal, and operational tasks, with average task completion times of 8-10 minutes [5] - The app incorporates a "double-check" mechanism to enhance content accuracy by allowing users to verify key data or conclusions through third-party agents [5] Group 4: Ecosystem Integration - Qianwen App has connected with various services within Alibaba's ecosystem, including Taobao, Amap, and Fliggy, supporting over 400 service capabilities, from food ordering to report writing and hotel booking [5]

Core Insights - The development of intelligent agents has accelerated significantly at the beginning of 2026, with notable advancements from companies like Anthropic and Alibaba [1][11] - Anthropic's release of Cowork aims to revolutionize the workplace by integrating large models with intelligent agent capabilities for general users, not just programmers [1] - Alibaba's Qianwen App has introduced a new AI Agent feature called "Task Assistant," which integrates with Alibaba's ecosystem to offer over 400 new functionalities for free [2][4] Group 1 - The Qianwen App can automate tasks such as ordering food by simply stating preferences, streamlining the entire process from selection to payment [5][20] - Users can consult the Task Assistant for shopping decisions, which can provide recommendations and direct links to payment [7][9] - The Task Assistant has demonstrated its ability to handle complex tasks like multi-brand group purchases, significantly reducing the time and effort required for users [12][18] Group 2 - The Task Assistant can create detailed travel plans, such as a two-day itinerary for a trip to Weihai, by analyzing user needs and sourcing information from various platforms [22][27] - The assistant integrates with Alibaba's services, allowing users to navigate, book tickets, and manage travel logistics seamlessly [29] - The interaction model has shifted from dialogue with a large model to task delegation to an intelligent agent, marking a significant evolution in user experience [31] Group 3 - Qianwen's Task Assistant is built on a new universal agent system that enhances task execution efficiency and accuracy through a hierarchical planning approach [33] - The system allows for continuous learning and improvement, enabling agents to refine their capabilities based on past experiences [35] - The integration of AI coding capabilities allows the assistant to autonomously generate tools for less common tasks, enhancing its functionality [36] Group 4 - The AI sector is entering a product explosion phase, with new offerings from various companies, including Anthropic and OpenAI, indicating a rapid evolution in intelligent agent applications [38] - Qianwen's launch is compared to the introduction of the first iPhone, suggesting it could signify a transformative moment in the AI landscape [38] - The shift from AI as a distant entity to a practical assistant in daily tasks represents a pivotal change in human-machine interaction [38]

Group 1 - The core asset of Hong Kong's AI sector, the Hong Kong Internet ETF (513770), experienced a short-term decline of 1.89%, indicating active buying interest during price dips [1] - Over the past 20 days, the Hong Kong Internet ETF (513770) has attracted over 1.8 billion yuan in capital [1] - Southbound funds have recorded a historic net inflow of 1,404.844 billion HKD into Hong Kong stocks for the year 2025, with a cumulative net inflow of 44.162 billion HKD as of January 14, 2026, showing a strong momentum in increasing positions in Hong Kong stocks [1] Group 2 - According to Guotai Junan Securities, AI applications are expected to transition from usable to highly usable by 2026, with diverse business models becoming established, positioning AI applications as a core theme in the AI industry market for that year [1] - The Hong Kong Internet ETF (513770) and its linked fund (017125) passively track the CSI Hong Kong Internet Index, heavily investing in leading internet companies like Alibaba-W and Tencent Holdings, with the top 10 holdings accounting for over 76% of the total [1] - The latest fund size of the Hong Kong Internet ETF (513770) exceeds 14.8 billion yuan, nearing the 15 billion yuan mark, setting a new record for fund size, and it supports T+0 trading, enhancing liquidity [1] Group 3 - The fund manager of the Hong Kong Internet ETF (513770) noted that while the Hong Kong market was disappointing in Q4 2025, this could lead to better performance in 2026 [2] - As of January, the daily trading volume in Hong Kong was less than 300 billion HKD, compared to 3 trillion HKD in A-shares, suggesting that a 10% capital outflow from A-shares to Hong Kong could double the trading activity in Hong Kong [2] - The current situation in the Hong Kong market is seen as lacking a logical catalyst, and once this logic is corrected, a rapid valuation increase is anticipated [2]

Core Viewpoint - Zhizhu (02513) saw a significant increase of over 6%, currently trading at 229.8 HKD with a transaction volume of 335 million HKD, following the announcement of a collaboration with Huawei on the open-source next-generation image generation model GLM-Image [1] Group 1: Company Developments - Zhizhu announced the launch of GLM-Image, the first state-of-the-art (SOTA) multimodal model fully trained on domestic chips, utilizing the Ascend Atlas 800T A2 device and MindSpore AI framework [1] - The GLM-Image model integrates image generation with language models, allowing for image generation at a cost of only 0.1 yuan per image when using API calls [1] Group 2: Market Outlook - Dongwu Securities views Zhizhu as a pure large model player benefiting from cloud-scale effects and the advantages of agent/programming scenarios [1] - The company is expected to leverage its strengths in local large model technology, open-source ecosystem development, and localized implementation capabilities in government and enterprise sectors [1] - There is a positive outlook for Zhizhu as the Chinese large model industry transitions from localized deployment to cloud services, indicating a long-term growth trend [1]

Core Viewpoint - The company Zhipu (02513) has seen a stock price increase of over 6%, currently trading at 229.8 HKD, with a transaction volume of 335 million HKD, following the announcement of a collaboration with Huawei on a new open-source image generation model, GLM-Image [1] Group 1: Company Developments - Zhipu announced the launch of GLM-Image, a next-generation image generation model developed in collaboration with Huawei, which is the first SOTA multimodal model fully trained on domestic chips [1] - The model utilizes the Ascend Atlas 800T A2 device and the MindSpore AI framework, completing the entire process from data to training [1] - GLM-Image integrates image generation with language models, allowing for image generation at a cost of only 0.1 yuan per image when using API calls [1] Group 2: Market Outlook - Dongwu Securities expresses optimism about Zhipu as a pure large model player, benefiting from cloud-scale effects and the advantages of Agent/programming scenarios [1] - The company is expected to leverage its strengths in local large model technology, open-source ecosystem layout, and localized implementation capabilities in government and enterprise sectors [1] - There is a long-term trend anticipated in the Chinese large model industry, shifting from localized deployment to cloud services, which is expected to benefit Zhipu [1]