Group 1 - The China Securities Software Service Index fell by 1.31% on January 27, 2026, with mixed performance among constituent stocks, led by Hehe Information up 1.07%, and followed by Shenxinfu up 0.86% and Hengsheng Electronics up 0.84%, while Guanghuan Xinwang led the decline [1] - The release of the 2.0 version of the KOUZI tool on January 19 is seen as a significant enhancement in capabilities such as Agent Skills, Agent Plan, and Agent Coding, addressing key pain points of general AI tools being business-ignorant and customized tools being costly and time-consuming, thus facilitating the transition of AI applications in enterprises from pilot projects to large-scale penetration [1] - CITIC Construction Investment Securities noted that the Agent industry is accelerating its evolution from single intelligence to multi-agent systems (MAS), with products like Anthropic CoWork and MiniMax Agent 2.0 deeply integrating local development and office workflows, upgrading AI roles from conversational assistants to digital employees with long-term planning and autonomous execution capabilities [1] Group 2 - As of December 31, 2025, the top ten weighted stocks in the China Securities Software Service Index include iFLYTEK, Kingsoft Office, Tonghuashun, Zhinancun, Hengsheng Electronics, Tuo Wei Information, Runhe Software, 360, Softcom Power, and Shenxinfu, collectively accounting for 60.89% of the index [2] - The Software ETF (159852) tracks the China Securities Software Service Index, serving as a convenient tool for capitalizing on opportunities in the computer software industry [2] - Investors can also access AI software investment opportunities through the Software ETF linked fund (012620) [2]

Core Insights - The article discusses the emergence of multi-agent systems (MAS) in the Agentic AI era, emphasizing the shift from individual models to collaborative problem-solving among AI agents [2][5] - A new framework called LatentMAS is introduced, which allows agents to collaborate in latent space rather than through traditional text communication, enhancing efficiency and performance [5][14] Group 1: LatentMAS Framework - LatentMAS enables agents to exchange internal hidden layer representations and KV-cache working memory, resulting in higher performance and reduced token usage [5][10] - The framework is designed to support richer latent reasoning and lossless communication between agents, significantly lowering computational complexity compared to text-based MAS [15][16] Group 2: Experimental Results - Comprehensive experiments on nine benchmark tasks show that LatentMAS outperforms both single models and text-based MAS, with accuracy improvements of up to 14.6% and token usage reductions of 70.8% to 83.7% [6][20][22] - LatentMAS achieves end-to-end reasoning speed increases of 4× to 4.3× compared to traditional methods, demonstrating its efficiency [21][25] Group 3: Efficiency and Performance - The framework allows for complex reasoning processes while significantly reducing the number of tokens used, achieving higher accuracy with fewer output tokens [28][29] - LatentMAS can provide additional speed improvements of 2.6× to 7× over text-based MAS, even when the latter is optimized with vLLM services [25][28] Group 4: Semantic Richness - The latent representations generated by LatentMAS are shown to be semantically rich and diverse, surpassing the expressiveness of discrete tokens used in text-based systems [30][31] - The study indicates that the potential reasoning captured in LatentMAS is not only effective but also contains more nuanced internal representations compared to traditional methods [31][32]

Core Viewpoint - The article emphasizes that 2026 will be a pivotal year for technology leaders, with unprecedented speed in transformation, innovation, and risk driven by artificial intelligence (AI) and a highly interconnected world [2]. Group 1: AI Supercomputing Platforms - AI supercomputing platforms integrate various computing paradigms to manage complex workloads, enhancing performance and innovation potential [5]. - By 2028, over 40% of leading companies will adopt hybrid computing architectures for critical business processes, a significant increase from the current 8% [6]. - The technology is already driving innovation across industries, significantly reducing drug modeling time in biotech and lowering portfolio risks in financial services [7]. Group 2: Multi-Agent Systems - Multi-agent systems consist of multiple AI agents that interact to achieve complex individual or collective goals, enhancing automation and collaboration [9]. - These systems allow for modular design, improving efficiency and adaptability in business processes [9]. Group 3: Domain-Specific Language Models (DSLM) - DSLMs are trained on specialized datasets for specific industries, providing higher accuracy and compliance compared to generic large language models (LLMs) [11]. - By 2028, over half of generative AI models used by enterprises will be domain-specific [12]. - Context is crucial for the success of AI agents based on DSLMs, enabling them to make informed decisions even in unfamiliar scenarios [13]. Group 4: AI Security Platforms - AI security platforms provide unified protection mechanisms for third-party and custom AI applications, helping organizations monitor AI activities and enforce usage policies [13]. - By 2028, over 50% of enterprises will utilize AI security platforms to safeguard their AI investments [15]. Group 5: AI-Native Development Platforms - AI-native development platforms enable rapid software development, allowing non-technical experts to create applications with AI assistance [17]. - By 2030, 80% of enterprises will transform large software engineering teams into smaller, more agile teams empowered by AI [17]. Group 6: Confidential Computing - Confidential computing reshapes how enterprises handle sensitive data by isolating workloads in trusted execution environments [18]. - By 2029, over 75% of business workloads processed in untrusted environments will be secured through confidential computing [18]. Group 7: Physical AI - Physical AI empowers machines and devices with perception, decision-making, and action capabilities, providing significant benefits in automation and safety-critical industries [19]. Group 8: Proactive Cybersecurity - Proactive cybersecurity is becoming a trend as organizations face increasing threats, with predictions that by 2030, proactive defense solutions will account for half of enterprise security spending [23]. Group 9: Geopolitical Data Migration - Geopolitical risks are prompting companies to migrate data and applications to sovereign or regional cloud services, enhancing control over data residency and compliance [26]. - By 2030, over 75% of enterprises in Europe and the Middle East will migrate virtual workloads to solutions that mitigate geopolitical risks, up from less than 5% in 2025 [26].

Core Viewpoint - The article discusses the evolution and differentiation between AI Agents and Agentic AI, highlighting their respective roles in automating tasks and collaborating on complex objectives, with a focus on the advancements since the introduction of ChatGPT in November 2022 [2][10][57]. Group 1: Evolution of AI Technology - The development of AI technology has progressed from early expert systems like MYCIN to modern AI Agents and Agentic AI, marking a significant paradigm shift in capabilities [10][11]. - ChatGPT's release in November 2022 is identified as a pivotal moment that catalyzed the evolution of AI Agents, transitioning from passive responders to more autonomous systems capable of executing multi-step tasks [12][24]. - The introduction of frameworks like AutoGPT and BabyAGI in 2023 signifies the formal establishment of AI Agents, which integrate LLMs with external tools to perform complex tasks [12][24]. Group 2: Characteristics of AI Agents - AI Agents are defined as modular systems driven by LLMs and LIMs, designed for task automation, filling the gap where generative AI lacks execution capabilities [13][16]. - Three core features distinguish AI Agents from traditional automation scripts: autonomy, task-specificity, and reactivity [16][17]. - The integration of tools allows AI Agents to overcome limitations of static knowledge and hallucination issues, enabling them to perform real-time data retrieval and processing [19][20]. Group 3: Agentic AI and Multi-Agent Collaboration - Agentic AI represents a shift towards multi-agent collaboration, where multiple AI Agents work together to achieve complex goals, enhancing system-level intelligence [24][27]. - The architecture of Agentic AI includes dynamic task decomposition and shared memory, facilitating efficient collaboration among specialized agents [33][36]. - Real-world applications of Agentic AI demonstrate its advantages in various fields, such as healthcare and agriculture, where multiple agents coordinate to optimize processes [37][38]. Group 4: Challenges and Future Directions - Both AI Agents and Agentic AI face challenges, including causal reasoning deficits and coordination issues among multiple agents [48][50]. - Proposed solutions include enhancing retrieval-augmented generation (RAG), implementing causal modeling, and establishing shared memory architectures to improve collaboration and decision-making [49][53]. - The future roadmap emphasizes the need for deeper causal reasoning, transparency in decision-making, and ethical governance to ensure the responsible deployment of AI technologies [56][59].

Core Viewpoint - The article discusses the introduction of MAS-GPT, a new generative design paradigm for Multi-Agent Systems (MAS), which simplifies the process of creating MAS to a single query input, making it as easy as interacting with ChatGPT [2][9]. Group 1: Introduction of MAS-GPT - MAS-GPT is a collaborative effort from institutions like Shanghai Jiao Tong University and Oxford University, aiming to facilitate the development of MAS as a step towards achieving Artificial General Intelligence (AGI) [2][3]. - The system allows users to generate a complete and executable MAS with just one query, significantly streamlining the process [2][12]. Group 2: Challenges in Existing MAS Methods - Current MAS methods face three fundamental issues: lack of adaptability, high costs, and low generalization capabilities, which hinder their widespread application [5][7]. - Existing systems require extensive manual input and multiple rounds of LLM calls, making them inefficient and costly [7]. Group 3: MAS-GPT's Solution - MAS-GPT transforms the design of MAS into a language generation task, allowing for the automatic generation of MAS from user queries [9][10]. - The generated MAS is presented in Python code, eliminating the need for manual coding [9]. Group 4: Performance and Evaluation - MAS-GPT has been tested against over ten existing methods across eight benchmark tasks and five mainstream models, demonstrating superior performance [16]. - It achieved an average accuracy improvement of 3.89% over the strongest baseline and maintained robust performance on unseen tasks [17]. Group 5: Cost Efficiency and Compatibility - MAS-GPT operates at nearly half the inference cost compared to other systems like DyLAN and GPTSwarm while delivering better results [18]. - The MAS generated by MAS-GPT shows strong compatibility and consistent performance across different LLMs [20]. Group 6: Future Potential and Community Engagement - MAS-GPT has significant potential for future development, with the ability to generate novel MAS structures and adapt to new tasks [24][25]. - The MASWorks community aims to connect researchers globally, fostering collaboration and knowledge sharing in the MAS field [30][31].

本文由上海人工智能实验室，悉尼大学，牛津大学联合完成。第一作者周恒为上海 ailab 实习生和 Independent Researcher 耿鹤嘉。通讯作者为上海人工智能实验 室青年科学家白磊和牛津大学访问学者，悉尼大学博士生尹榛菲，团队其他成员还有 ailab 实习生薛翔元。 ReSo 框架（ Re ward-driven & S elf- o rganizing）为复杂推理任务中的多智能体系统（MAS）提供了全新解法，在处理复杂任务时，先分解生成任务图，再为每个 子任务匹配最佳 agent。将任务图生成与奖励驱动的两阶段智能体选择过程相结合，该方法不仅提升了多智能体协作的效率，还为增强多智能体的推理能力开辟了 新路径。 研究背景：LLM 推理能力的掣肘与突破口 近年来， 增加推理时间（Inference Time Scaling） 被广泛认为是提升大语言模型（Large Language Models, LLMs）推理能力的重要途径之一。一方面，通过在训 练后阶段引入强化学习与奖励模型，可优化单一模型的推理路径，使其在回答前生成中间步骤，表现出更强的逻辑链构建能力；另一方面，也有研究尝试构建 多 智能体 ...

锌财经 . 以下文章来源于锌财经 ，作者路世明 关注新商业，关注新经济。一家由人民网投资的数字化媒体机构。 将投中网设为"星标⭐"，第一时间收获最新推送 AI Agent的竞争既充满想象力，又暗藏风险。 作者丨路世明 编辑丨 大风 来源丨锌财经 Manus的出现，激起了科技与资本市场的双重震荡，一时间AI Agent相关概念股集体大涨，阿里、谷 歌、微软等科技巨头密集发布智能体研发计划...... 而在这场热潮的背后，是AI技术从"被动应答"向"主动执行"的范式跃迁。 尽管市场的评价褒贬不一，但不能否认，Manus的突破性在于，它首次验证了通用型AI Agent在复杂 场景下的商业化可行性。 传统的大语言模型虽能生成文本，却难以闭环执行任务，而Manus通过"规划-验证-执行"的架构，将 AI大模型的认知能力转化为生产力工具。 根据麦肯锡等多份权威报告，在多元化需求驱动下，AI Agent市场呈爆发式增长态势，2024年全球 AI Agent市场规模约为51亿美元，预计2030年将飙升至471亿美元，复合年增长率高达44.8%。 然而，这场"智能体浪潮"并非坦途。技术瓶颈与商业野心的碰撞，让AI Agent的 ...