AI Agent

Search documents
小易智联董事长杨振:XYZ模型可帮助企业摆脱对高性能显卡的依赖
Sou Hu Cai Jing· 2025-09-19 06:13
中国商报(记者 赵熠如 文/图)9月15日,重庆小易智联智能技术有限公司(以下简称小易智联)董事长杨振在接受中国商报记者采访时表示,小易智联研 发的XYZ端侧模型可帮助企业摆脱对高性能显卡的依赖。 重庆小易智联智能技术有限公司董事长杨振。 据介绍,XYZ端侧模型是小易智联全自研、全正向、全国产的AI模型。它凭借创新架构,突破"卡脖子"难关,摆脱对高性能显卡的依赖,具备低资源需求、 低成本部署特性,可广泛应用于多场景。XYZ模型的整体大小仅10MB,比传统AI模型缩小99%以上;耗电量仅0.008度/小时,部署成本降低99%以上。该模 型实现从芯片到算法的全链条国产化,所有数据处理100%本地化运行,为关键基础设施和敏感行业提供了安全可靠的AI解决方案。 "经过中国信通院的代码审核,我们的国产化自研率已达到99.66%,意味着这是一个完全由我们一行一行的代码构建起来的基座模型。"杨振说。 "当前提倡'人工智能+制造',其目标是打造'黑灯工厂',但企业自建算力中心、更换生产线设备的成本极高。"杨振说,"XYZ模型可显著降低成本,帮助企 业在算力层面摆脱对GPU的依赖,突破国际算力垄断。XYZ模型仅靠CPU即可完成 ...
港股异动丨银盛数惠逆势大涨10% 推出首款AI Agent产品“数惠助手”
Ge Long Hui· 2025-09-18 06:58
消息面上,银盛数惠近日推出了其首款AI Agent产品——"数惠助手",同时其亦为国内数字权益消费场景下首个AI智能体。"数惠助手"聚焦本地生活服务赛 道,"找优惠,问数惠!",将全力打造AI智能优惠生活的入口,未来战略蓝图是要推动实现从工具到综合服务平台的飞跃性跨越。在本地生活赛道率先迈入 AI智能体转型升级的背景下,公司在AI Agent布局上的先行一步,不仅有助于抢占市场份额,更能积累宝贵的用户数据和运营经验,在未来的行业整合中占 据有利位置。(格隆汇) 港股三大指数今日集体下跌,银盛数惠(3773.HK)逆势走高,一度涨10.07%至1.64港元,市值达6.8亿港元。 ...
独家丨前安克 UV 打印机产品负责人付瀚龙计划创业,多家头部机构火速跟进
雷峰网· 2025-09-18 00:24
付瀚龙于2023年9月加入安克,曾短暂担任UV打印机负责人。据了解,安克UV打印机的产品定义由付瀚龙完 成,他也算得上是立下了汗马功劳。 以下文章来自雷峰网子账号【鲸犀】, 想了解更多独家或干货可点击关注: " 付瀚龙先后在大疆、元戎启行、安克待过。 " 作者丨姚单 编辑丨余快 据 雷峰网·鲸犀 近日报道,前安克UV打印机产品负责人付瀚龙从安克离职后,计划创业。该项目正在如火如 荼地融资中,多家头部机构有意参与投资。 据知情人士透露,付瀚龙的项目在一级市场非常火热,有望直接完成多轮融资。 // 近期热门文章 独家丨前钉钉CEO叶军计划创业,投身于AI Agent赛道 原大疆 Mavic 3 产品经理李进吉创业,瞄准影像眼镜赛道 发力UV打印机,创想三维IPO前景如何? 加入安克前,付瀚龙先后在大疆、元戎启行待过。付瀚龙在2017年8月加入大疆担任高级产品工程师,并于 2021年4月离职后加入元戎启行担任高级产品经理。 付瀚龙此番创业,仍在寻找方向当中,并同步搭建团队。 多方消息人士透露,付瀚龙曾向3D影像创业者探讨影像产业链情况,但最终未确定是否投身于3D影像行业创 业。 不过,3D影像被视为不是一个很好的创 ...
Shopify 经验贴:如何搞出一个生产级别可用的 AI Agent 系统?
Founder Park· 2025-09-17 12:50
Core Insights - Shopify's experience in developing the AI assistant Sidekick highlights the evolution from a simple tool to a complex AI agent platform, emphasizing the importance of architecture, evaluation methods, and training techniques [2][4]. Group 1: Evolution of Sidekick Architecture - The core of Sidekick is built around the "agentic loop," where human input is processed by a large language model (LLM), actions are executed, feedback is collected, and the cycle continues until the task is completed [5]. - Simplifying architecture and ensuring tools have clear boundaries are crucial for effective design [6]. - The challenge of tool complexity arose as the functionality expanded, leading to the "Death by a Thousand Instructions" problem, which hindered system speed and maintenance [10][12]. Group 2: Evaluation System for LLMs - A robust evaluation system is essential for deploying intelligent agent systems, as traditional software testing methods are inadequate for the probabilistic outputs of LLMs [17]. - The shift from "golden datasets" to "Ground Truth Sets" reflects a focus on real-world data distribution, enhancing the relevance of evaluation standards [20]. - The process includes aligning LLM judges with human evaluations, improving correlation from 0.02 to 0.61, close to human benchmarks [21]. Group 3: Training and Reward Mechanisms - The Group Relative Policy Optimization (GRPO) method was adopted for model fine-tuning, utilizing LLM judges as reward signals [31]. - The issue of "reward hacking" was identified, where models exploited the reward system, necessitating updates to both syntax validators and LLM judges [32][34]. - Iterative improvements were made to address these challenges, ensuring a more reliable training process [34]. Group 4: Key Recommendations for Building AI Agent Systems - Maintain simplicity and resist the temptation to add tools without clear boundaries, prioritizing quality over quantity [37]. - Start with modular designs like "Just-in-Time Instructions" to maintain understandability as the system scales [37]. - Anticipate reward hacking and build detection mechanisms early in the development process [37].
LLM开源2.0大洗牌:60个出局,39个上桌,AI Coding疯魔,TensorFlow已死
3 6 Ke· 2025-09-17 08:57
Core Insights - Ant Group's open-source team unveiled the 2.0 version of the "2025 Large Model Open Source Development Ecosystem Panorama" at the Shanghai Bund Conference, showcasing significant changes in the open-source landscape [2][4][10] Group 1: Ecosystem Changes - The updated panorama includes 114 projects, a decrease of 21 from the previous version, with 39 new projects and 60 projects that have exited the stage, including notable ones like TensorFlow, which has been overtaken by PyTorch [4][5] - The overall trend indicates a significant reshuffling within the ecosystem, with a median age of only 30 months for projects, highlighting a youthful and rapidly evolving environment [5][10] - Since the "GPT moment" in October 2022, 62% of the projects have emerged, indicating a dynamic influx of new entrants and exits [5][10] Group 2: Project Performance - The top ten most active open-source projects reflect a focus on AI, LLM, Agent, and Data, indicating the primary areas of interest within the ecosystem [7][9] - The classification framework has evolved from broad categories to more specific segments, including AI Agent, AI Infra, and AI Data, emphasizing the shift towards an "agent-centric" era [10][19] Group 3: Contributions by Region - Among 366,521 developers, the US and China contribute over 55%, with the US leading at 37.41% [10][12] - In specific areas, the US shows a significant advantage in AI Infra and AI Data, with contributions of 43.39% and 35.76% respectively, compared to China's 22.03% and 21.5% [12][14] Group 4: Methodological Evolution - The methodology for selecting projects has shifted from a known starting point to a broader approach that captures high-activity projects, increasing the threshold for inclusion [15][18] - The new methodology aligns with Ant Group's goal of providing insights for internal decision-making and guidance for the open-source community [15][18] Group 5: AI Agent Developments - The AI Agent category has evolved into a structured system with various specialized tools, indicating a transition from chaotic growth to systematic differentiation [19][21] - AI Coding has expanded its capabilities, covering the entire development lifecycle and supporting multimodal and context-aware functionalities [23][27] Group 6: Market Trends - The report predicts significant commercial potential in AI Coding, with new revenue models emerging from subscription services and value-added features [24][27] - Chatbot applications have seen a peak but are now stabilizing, with a shift towards integrating knowledge management for long-term productivity [28][30] Group 7: Infrastructure and Operations - The Model Serving segment remains a key battleground, with high-performance cloud inference solutions like vLLM and SGLang leading the way [42][45] - LLMOps is rapidly growing, focusing on the full lifecycle management of models, emphasizing stability and observability [50][52] Group 8: Data Ecosystem - The AI Data sector appears stable, with many projects originating from the AI 1.0 era, but is facing challenges in innovation and engagement [58][60] - The evolution of data infrastructure is anticipated, moving from static repositories to dynamic systems that provide real-time insights for models [60][61] Group 9: Open Source Dynamics - A trend towards customized open-source licenses is emerging, allowing for more control and flexibility in commercial negotiations [62][63] - The landscape of open-source projects is being challenged, with some projects operating under restrictive licenses, raising questions about the definition of "open source" [62][63] Group 10: Competitive Landscape - The competitive landscape is marked by a divergence between open-source and closed-source models, with Chinese projects flourishing while Western firms tighten their open-source strategies [67][68] - The introduction of MoE architectures and advancements in reasoning capabilities are becoming standard features in new models, indicating a shift in focus from scale to reasoning [69][70]
李开复:法律行业是AI Agent落地“黄金赛道”,企业法务正迈入新阶段
Xin Lang Ke Ji· 2025-09-17 03:20
第二,支持个性化定制法务智能体,让AI更懂企业业务,完成更高质量交付。既提供通用的法务智能 体,也可以帮助企业梳理内部知识库,进行智能体的个性化定制,在企业的语境中更快执行具体任务。 第三,支持跨部门/子公司便捷管理和分配订阅的智能体,提供全周期智能体更新与维护服务。通过法 务智能体平台,企业可自由分配和订阅所需智能体,享受智能体的后期更新与维护服务,确保数据库的 即时性。(文猛) 新浪科技讯 9月17日上午消息,近日,法天使联合零一万物发布法务智能体平台,宣布将以"法务 +AI"更快地帮行业提高效率,带来效率和营收的规模化增长。零一万物创始人兼CEO李开复为发布会 致辞称:"AI Agent已从'可用'进化为'好用'的智能决策Agent,真正成为企业生产力,法律行业凭借扎实 的数字化基础,成为生成式AI与AI Agent快速落地、深度赋能的'黄金赛道'。" 责任编辑:江钰涵 "此次零一万物与法天使联合发布'法务智能体平台',不仅实现了AI Agent与法律实务的深度融合,更标 志着企业法务正式迈入AI数智化新阶段。"李开复表示。据悉,此次发布的"法务智能体平台"主要有三 项功能亮点: 第一,支持企业本地化部 ...
8点1氪:特斯拉“车顶维权”女车主胜诉;太二酸菜鱼客服回应“活鱼现做”质疑;迪卡侬客服回应“广告词采用‘处女地’一词”争议
36氪· 2025-09-17 00:09
Group 1 - The core viewpoint of the article is that the "car roof rights protection" female car owner, Zhang Yazhou, has won a court ruling against Tesla, requiring the company to provide complete driving data from the 30 minutes prior to a traffic accident [4] - The Beijing Daxing District People's Court ruled on September 16, 2023, that Tesla must provide the requested data, which is crucial for determining the cause of the accident that occurred in February 2021 [4] - Zhang expressed her relief and determination to seek further judicial appraisal based on the data provided, marking a significant moment in her ongoing legal battle with Tesla [4] Group 2 - Taier Suancaiyu's customer service confirmed that currently, there are 68 "fresh live stores" operating nationwide, with only one in Hangzhou using live fish for preparation [5] - The company has faced scrutiny regarding its claims of using live fish, as a recent test showed that multiple dishes were served within seven minutes, raising questions about the freshness of the ingredients [5] - Taier Suancaiyu has also been reported to be closing stores in various cities for upgrades, indicating a potential shift in their operational strategy [7] Group 3 - Anta Group reported that 46 employees, including one at the executive level, have been transferred to judicial authorities for suspected illegal activities, highlighting ongoing issues with corporate governance [8] - The company has been actively addressing corruption and misconduct, with a total of 74 employees dismissed for serious violations since the beginning of 2025 [8] Group 4 - The price of spot gold reached a historic high of $3,690 per ounce on September 16, 2023, leading to a surge in domestic gold jewelry prices, with some brands exceeding 1,090 yuan per gram [11] - This increase in gold prices reflects broader trends in the commodities market and may influence consumer behavior in the jewelry sector [11] Group 5 - HeSai Technology officially listed on the Hong Kong Stock Exchange on September 16, 2023, raising approximately HKD 4.16 billion, marking the largest IPO in the global lidar industry to date [14] - The stock opened at HKD 234, reflecting a 9.96% increase on its first day of trading, indicating strong investor interest [14]
中国信通院云大所所长何宝宏:数字原生,点亮未来智能化社会
Sou Hu Cai Jing· 2025-09-16 20:44
"数字原生推动企业数字化从被动转型走向主动原生,彻底地释放AI潜力。 在AI大模型规模持续扩大、应用场景日益丰富的当下,行业却普遍面临一大关键难题:如何跳出同质化竞争与低水平复制的困局,让AI技术真正转化为 业务与社会价值,推动产业数智化转型迈向更深层次? 9月8日,由数智猿与数据猿联合主办、中关村科学城公司协办,并获新华社中国经济信息社等多家机构支持的"2025第五届数智化转型升级发展论坛—— 暨AI大模型&AI Agent趋势论坛"针对这些问题,展开了深入而广泛的讨论。 何宝宏博士说,如今的AI时代,正以高度相似的逻辑复刻这一进程。Transformer架构如同当年的TCP/IP,一统人工智能技术江湖,成为AI基础设施的核 心;以算力为核心的英伟达等企业,也正因这一基础设施的需求爆发,迎来估值的爆发式增长。 生态层的突破同样清晰。OpenAI的ChatGPT,恰似当年的网景浏览器,以具象化的应用场景激活了AI的技术潜力,点燃了人工智能革命的火种。 应用层面,AI也在重复从文字到多模态的演进,从通用大模型向私域部署、垂直领域专用模型延伸。只不过,互联网时代的"流量",如今被AI时代 的"Token"所替代, ...
X @Decrypt
Decrypt· 2025-09-16 17:14
Google Reveals AI Agent Payments Protocol Backed by Coinbase, Ethereum Foundation► https://t.co/RJ9YVA6vRn https://t.co/RJ9YVA6vRn ...
第四范式(6682.HK):营收大超预期 先知AI平台高速增长
Ge Long Hui· 2025-09-16 10:47
Core Insights - Fourth Paradigm reported a significant revenue increase of 40.7% year-on-year for H1 2025, reaching 2.626 billion yuan, with a gross profit of 990 million yuan, reflecting a 25.4% growth [1] - The company's AI platform, "Prophet AI," generated 2.149 billion yuan in revenue, marking a 71.9% increase and accounting for 81.8% of total revenue [1][2] - The average revenue per benchmark user increased by 56.6% year-on-year to 17.98 million yuan, indicating enhanced value capture in the enterprise sector [1] Revenue and Profitability - The company achieved a gross margin of 37.71%, down 4.61 percentage points from the previous year, primarily due to a shift towards integrated delivery models [1][2] - Adjusted net loss narrowed by 71.2% to 44 million yuan, showcasing improved financial performance despite the loss [1] R&D and Expenses - R&D expenses rose by 5.1% year-on-year to 893 million yuan, with an R&D expense ratio of 34.0%, down 11.5 percentage points [1][2] - Total expenses for sales, management, and R&D were 189 million, 86 million, and 893 million yuan respectively, with corresponding expense ratios of 7.2%, 3.29%, and 34.0% [2] Market Position and Future Outlook - Fourth Paradigm has maintained the largest market share in China's machine learning platform sector for seven consecutive years, indicating strong competitive positioning [3] - The company is expected to benefit from the digital transformation across various industries in China, with projected revenues of 6.937 billion, 9.062 billion, and 11.797 billion yuan for 2025-2027, reflecting a compound annual growth rate of approximately 30% [3]