一则消息突然引爆。 今日，小米系集体爆发，小米集团一度大涨超5%，金山云一度大涨超18%。消息面上，小米大模型微信公众号称，小米开源首个为推理 （Reasoning）而生的大模型"Xiaomi MiMo"，联动预训练到后训练，全面提升推理能力。 当前，AI大模型竞争再度进入白热化。4月29日，阿里巴巴通义千问发布了新版Qwen 3系列模型。同一天，世界首富马斯克也宣布，下周Grok 3.5早 期测试版将向SuperGrok订阅者发布。 券商机构分析认为，国内外大模型能力差距缩小，国产算力芯片持续进化，国内头部互联网厂商加大Capex投入，为国内AI突围之路奠定了基石。伴 随国内外巨头大模型持续迭代，C端或有望迎来杀手级应用。 全线大涨 4月30日，小米系全线拉升，小米集团盘中一度大涨超5%，金山云一度大涨超18%，金山软件最高涨超6%。 与此同时，港股云计算板块持续拉升，截至午间收盘，金蝶国际涨4.7%，中国软件国际涨2.72%，阿里巴巴-W涨1.48% 据"小米大模型"公众号消息，今日，小米开源首个为推理（Reasoning）而生的大模型"Xiaomi MiMo"，联动预训练到后训练，全面提升推理能力。 据 ...

Group 1 - The Zhongzheng All-Index Software Development Index showed active performance with an intraday increase of 2.75% [1] - The Software ETF Fund experienced a strong performance, with an intraday increase of 3.23%, and several component stocks, including Creative Software, reached the daily limit, while Xinghuan Technology-U and Zhongke Star Map rose over 6% [1] - The 2024 annual performance briefing for the AI and software industry on the Sci-Tech Innovation Board will be held on April 29, with 23 companies participating, indicating positive developments in order situations driven by AI [1] Group 2 - Ping An Securities noted that the DeepSeek series of large models has gained widespread attention due to its open-source, low-cost, and high-performance characteristics, which will accelerate the application of large models in B-end and C-end scenarios [2] - The overall demand for computing power is expected to continue to rise, with inference and edge computing power anticipated to grow faster, driven by the adaptation of DeepSeek with domestic AI chips [2] - The computer industry’s 2024 performance forecast indicates that computer equipment will outperform software development and IT services, while the latter shows better performance in turning losses into profits, suggesting growth potential in the industry [2]

Group 1: Financial Performance - The company's AI comprehensive service revenue increased to 208 million in 2024, with a significant growth of 86% in Q4 of the previous year, achieving profitability [1][4] - In Q1 2025, the AI comprehensive service revenue showed a year-on-year increase, although the company experienced a loss [4][8] - The gross margin for AI comprehensive services in 2024 was 20.70% [4] Group 2: AI Business Development - The company aims to enhance its AI Agent capabilities, focusing on multi-task and complex task agents, with existing orders already in place [2] - The AI Agent product line includes various applications such as intelligent customer service agents and intelligent credit agents, enhancing operational efficiency in banking [2] - The company has developed a global distribution platform for AI content, including micro-short films, although these products currently contribute a small percentage to overall revenue [3] Group 3: Research and Development - R&D expenses for Q1 2025 were 46.47 million, a 22.77% increase from 37.85 million in the same period last year [8] - The primary focus of R&D investments includes multi-modal applications, AI Agents, and large language models [8] - The company has established a comprehensive AI service framework, covering computational infrastructure, algorithms, and multi-modal applications [7] Group 4: Strategic Partnerships - The company collaborates with Alibaba Cloud as a partner and service provider for AI large model frameworks, enhancing its capabilities in the financial sector [6] - It has formed extensive partnerships with leading enterprises in the AI field, promoting the application of AI technologies across various industries [7]

Core Viewpoint - Taxfriend Software Group Co., Ltd. reported significant revenue growth in Q1 2025, driven by its high-value transformation strategy and the accelerated promotion of compliance tax optimization products [1][2] Group 1: Financial Performance - The company achieved an operating income of 449 million yuan, representing a year-on-year increase of 23.72% [1] - The net profit attributable to shareholders was 24.74 million yuan during the same period [1] Group 2: High-Value Transformation - Taxfriend's transformation focuses on creating an AIBM digital ecosystem in the smart tax and finance sector, leveraging Agentic AI technology to enhance compliance and financial management services [1] - The company aims to capture the trillion-yuan compliance tax optimization market through a joint operation and business management model [1] Group 3: Product Development and Market Strategy - The rapid growth in user adoption of basic compliance products like Risk Control Warning Agent and Compliance Assessment Agent contributed significantly to revenue growth [2] - Advanced compliance products such as Tax Optimization Planning Agent and Consulting Copilot are also experiencing increased user volume [2] - The "Xiyou" intelligent agent is designed to cover various application scenarios across different levels of tax authorities and government agencies, facilitating a shift towards high-value data services in digital governance [2] Group 4: Future Outlook - Despite increased short-term investments, the company anticipates that its high-value tax and finance services will continue to contribute to revenue and improve profitability [2] - The company is committed to building an Agent application factory that covers the entire tax and finance value chain, while also enhancing digital governance services driven by data [2]

Core Insights - The 8th Digital China Construction Summit was held in Fuzhou, Fujian Province, focusing on how digital technology is reshaping the financial ecosystem and the training of digital finance talents [1] - The forum featured prominent figures from academia and industry, discussing the transformative impact of AI and large models in finance [1] Group 1: AI and Financial Innovation - Zhang Wenbin, CTO of Du Xiaoman, highlighted the shift from generative models to reasoning models, emphasizing their enhanced capabilities in complex logical reasoning [3] - The application of reasoning models in finance has evolved from peripheral areas like customer service to core scenarios such as user experience and risk decision-making [3][4] - AI Agents are revolutionizing customer interaction by providing seamless online guidance and real-time responses, thus improving user experience and reducing reliance on manual processes [4] Group 2: Risk Management Enhancements - Traditional risk management processes often lead to information loss due to the transformation of raw data into structured variables; reasoning models can utilize full-dimensional raw data to enhance data efficiency [4] - Reasoning models can identify high-risk behaviors, such as suspicious transfers to high-risk accounts, by analyzing user transaction data [4] Group 3: Implementation Strategies for AI - Zhang Wenbin proposed starting with "small cuts" to build Agents, focusing on specific scenarios and customer segments to develop differentiated models [4] - The recommendation includes applying AI in real-world scenarios to generate data that can optimize models, creating a feedback loop of application, data accumulation, model iteration, and effect optimization [4] - Companies should concentrate computational power and talent to establish teams that accelerate AI application, prioritizing the cultivation of "AI-aware talents" to drive organizational transformation [4]

可能是更高效、更便宜的 Agent 实现路径。 文 丨 孙海宁 编辑 丨 程曼祺 主流 AI Agent 都把大语言模型（LLM，或者它的多模态版本）当作 "大脑"，靠一个或几个 LLM 编 排工作、调用工具。但也有另一条路：Agent 规划、作业靠不依赖自然语言的强化学习模型，LLM 只 充当 Agent 和人类的 "交互层"。 不一样的想法，来自去年 10 月成立，至今只有 4 个正式员工的 Pokee.ai。 Pokee.ai 创始人朱哲清有十余年强化学习研究、落地经验。2017 年起，从杜克大学计算机科学专业毕 业的朱哲清，一边在斯坦福大学攻读强化学习方向博士学位，师从 Benjamin Van Roy；一边在 Meta 工作，曾任 Meta"应用强化学习" 部门负责人，他用强化学习算法改善内容推荐系统，把上任前只剩 3 人，一度要关停的部门扩张至 10 余人，为 Meta 增收 5 亿美元。 靠 LLM 规划、决策，是个自然而主流的想法。OpenAI Operator 和网页交互、操作电脑的能力基于 GPT-4o 模型，Manus 完成任务则是靠 Claude 3.5 Sonnet 模型做长程规划。 ...

谈到这，也让人不禁思考，Agent成熟了吗？从技术发展历程来看，Agent从最初按固定规则回应，到如 今能自主决策、协作共事，经历了巨大飞跃。像2011年IBM Watson在智力问答节目中战胜人类选手以 及苹果Siri推出，标志着AI Agent进入成熟阶段；2022年ChatGPT问世，让AI Agent拥有自主执行复杂任 务能力 ，将其能力推向新高度。判断Agent成熟度可以从多个维度出发，比如上下文窗口提升让其能在 复杂任务保持连贯"工作记忆"；思维链与推理引擎发展，使其能非线性思考和自我修正；具备环境交互 能力，通过API调用等实际操作影响数字环境；实现多模态处理整合，能处理图像、声音等多种信息。 尽管当前Agent技术取得很大进展，但仍面临可靠性与稳定性、安全边界、隐私与数据安全、幻觉与错 误决策等技术挑战，以及责任归属、工作替代与转型等社会与伦理挑战 ，距离真正成熟或许还有一段 路要走，而千问3这样的模型进步，也可能会为Agent发展注入新动力，推动其不断完善成熟。 加入"中国IC独角兽联盟"，请点击进入 投稿 、 商务合作 请微信 dolphinjetta 是说芯语，欢迎关注分享 申请入围"中 ...

当AI Agent实现从回答问题到解决问题的跃升，人与AI Agent实现协同办公，甚至可以商量调整任 务步骤，会是一种怎样的工作体验？ 2025年，被称为AI Agent爆发元年，行业巨头正加速Agent落地。产品的应用正从人机交互升级为人 机协作，以金融行业为代表的有着大量复杂数据处理和深入研究需求的场景，正成为AI Agent应用 场景落地的热土。 近日，字节的 Agent 产品 "扣子空间（space.coze.cn） " 开启内测，产品聚焦于解决工作场景中的 复杂问题，在通用Agent基础上，引入了2个专家 Agent，其中切入金融工作场景的"华泰A股观察助 手"，上线后收获众多用户围观和测评。 "华泰A股观察助手" 由扣子团队与华泰证券联合探索孵化，这个专家Agent在扣子空间通用Agent技 术支持的基础上，还引入了华泰证券的专业能力加持，探索实现定制化信息整合。相较于目前市场上 同类产品，其着力简化了用户进行市场研究和交叉验证的流程。 据扣子团队介绍，扣子空间的长期目标是构建一个开放的Agent生态系统，通过MCP持续拓展Agent 能力边界，让用户与Agent实现高效协同，全面提升生产力。 ...

Summary of Key Points from the Conference Call Industry Overview - The conference call primarily discusses the AI industry, focusing on developments in AI applications and the performance of companies in the sector, particularly in the US and A-share markets [1][2][3][4]. Core Insights and Arguments - **US Market Performance**: The US stock market has seen a significant rebound in AI assets, with many stocks rising between 20% to 50%. Notable companies like Palantir have rebounded nearly 50%, while major players like Google have shown strong capital expenditure in computing power [2][3]. - **AI Application Commercialization**: Companies like Cursor and Midjourney have achieved substantial commercial success in a short time. ChatGPT is highlighted as the most profitable native AI application, with projected revenues exceeding $5 billion in 2025 [4][15][16]. - **AI Agent Development**: The biggest change in the AI industry in 2025 is the focus on application segments, with companies actively developing agent products to enhance engineering capabilities. The second half of the year is expected to see widespread application of multi-agent collaboration [1][6][11]. - **A-share Market Outlook**: The A-share market has shown average performance in AI-related stocks, with expectations of a rebound post-May Day, particularly for companies that performed well in Q1 [1][7]. Important but Overlooked Content - **Model Updates**: Significant updates from major model developers are anticipated in May and June, which may enhance reasoning capabilities and broaden applications in both general and specific scenarios [5][6]. - **Profitability Trends**: AI technology has significantly impacted corporate performance, with many companies showing improved profit margins due to internal efficiency gains and faster revenue generation from AI applications [32][33]. - **Emerging Business Models**: New business models are emerging in the AI industry, including SaaS subscription models for small and medium enterprises and open AI platforms for ISV partners, which are diversifying traditional project delivery methods [37]. Company-Specific Insights - **ChatGPT**: As of April 2025, ChatGPT has reached 20 million paid users, with annualized revenue expected to exceed $5 billion. The introduction of a universal agent and potential new subscription models are anticipated to further enhance its commercialization [16]. - **Cursor and GitHub Copilot**: Cursor has achieved a valuation of approximately $200 million, while GitHub Copilot has an ARR of $400 million, showcasing the rapid growth in the AI programming sector [17][18]. - **AI Search and Other Applications**: Companies like Perplexity are exploring various business models in the AI search domain, while Midjourney leads in the image generation sector with an annual revenue of around $300 million [19][21]. Future Trends - **AI Agent Evolution**: The development of intelligent agents is seen as a key trend for 2025, with companies focusing on enhancing their capabilities to solve problems autonomously and improve tool usage [28][29]. - **Market Dynamics**: The AI industry is witnessing a shift towards B2B applications, with companies like HanDe and DingJie showing strong performance in serving industrial clients, while consumer-oriented companies are lagging [34][35]. This summary encapsulates the key points discussed in the conference call, highlighting the current state and future outlook of the AI industry and its major players.

Investment Rating - The report initiates coverage with a "Buy" rating for the company, setting a target price of HKD 70.24 per share based on a 7.0x PS valuation for 2025 [4]. Core Viewpoints - The company is positioned as a leading player in the marketing and sales SaaS sector in China, with a diverse product matrix and strong growth potential driven by AI technology and market dynamics [2][3][4]. - The company has demonstrated a robust revenue growth trajectory, with a projected CAGR of 31.1% from 2020 to 2024, and an expected revenue of HKD 23.66 billion in 2025 [4][62]. - The company’s AI+SaaS product offerings are expected to enhance customer payment willingness and drive revenue growth, supported by a growing number of large clients and an expanding sales team [9][40][41]. Summary by Sections Investment Logic - The diversification of traffic sources is leading to a trend towards holistic marketing, increasing the demand for marketing and sales tools. The rationalization of investment in the SaaS sector is clearing out smaller players, providing more space for leading domestic firms [2]. - The company has served over 200,000 enterprises and achieved a revenue of HKD 840 million in 2024, with a market share of approximately 2.4%, making it the industry leader [2][22]. Business Model and AI Strategy - The company’s product matrix and business model closely resemble that of HubSpot, which has achieved significant revenue growth in the U.S. market. The company aims to replicate this success in China [3][50]. - The company has launched several AI-driven products, including the Tforce marketing model and the AI-Agentforce platform, which are expected to enhance its service offerings and market competitiveness [9][35]. Financial Projections and Valuation - Revenue projections for 2025, 2026, and 2027 are HKD 23.66 billion, HKD 30.22 billion, and HKD 37.08 billion, respectively, with year-on-year growth rates of 51.78%, 27.75%, and 22.71% [4][76]. - The adjusted net profit is expected to turn positive in 2024, with projections of HKD 0.70 billion, HKD 1.12 billion, and HKD 2.24 billion for the following years [4][70]. Product and Marketing Dimensions - The company’s AI+SaaS product matrix is continuously expanding, with 311 functional modules by the end of 2024, significantly enhancing customer engagement and revenue per client [9][33]. - The company is actively expanding its client base, particularly among government and large state-owned enterprises, which is expected to contribute significantly to revenue growth [9][40]. Competitive Landscape - The marketing and sales SaaS market in China remains fragmented, but the report anticipates a consolidation towards leading players as competition intensifies [2][23]. - The company is well-positioned to capture a larger market share due to its comprehensive product offerings and strong financial backing for R&D [2][26].