Core Viewpoint - DEEP Robotics showcases its advancements in embodied intelligent robots and practical applications at WAIC 2025, highlighting its leadership in technology breakthroughs and real-world applications [2] Group 1: Industry Application - DEEP Robotics creates a demonstration area at the exhibition to realistically replicate the electric power inspection scenario, showcasing the practical capabilities of its quadruped robots in industrial settings [3] - The simulated electric power inspection process is managed by DEEP Robotics' self-developed "Intelligent Inspection System," which maintains an accuracy rate of over 95% through intelligent recognition and self-correction technology [5] - The Intelligent Inspection System can autonomously conduct inspections and return even in the event of network disconnection, significantly enhancing inspection efficiency and quality across large areas [5] Group 2: Product Matrix - DEEP Robotics presents a diverse product matrix covering multiple scenarios, including the latest advancements in key technologies such as intelligent perception, interaction, and motion control, featuring products like the Jueying Lite3 and flagship quadruped robot Jueying X30 [6] - The company's products and solutions have been successfully implemented in over 600 industry application projects across 44 countries and regions globally, covering various fields such as construction surveying, industrial operation and maintenance, emergency firefighting, electric power inspection, security patrol, and education research [8] Group 3: Future Direction - DEEP Robotics aims to continue its focus on independent innovation and increase R&D investment to drive ongoing breakthroughs in core technologies, while also exploring industry needs to promote deeper applications of embodied intelligent robots [10] - The company is committed to contributing to a more efficient, safer, and smarter future across various industries through the large-scale application of its intelligent robots [10]

Core Viewpoint - The release of "Zhiyuan Lingqu OS" marks a significant advancement in the field of embodied intelligence, aiming to standardize and scale the industry through an open-source operating system framework [2][3][4]. Group 1: Overview of Zhiyuan Lingqu OS - "Zhiyuan Lingqu OS" is the first reference framework for embodied intelligent operating systems, facilitating a full link from hardware drivers to intelligent services [2][3]. - The operating system promotes a "layered open-source, co-construction and sharing" model, enhancing the existing high-performance middleware AimRT for stable and efficient distributed communication [2][4]. Group 2: Industry Challenges and Solutions - The robotics industry currently faces issues such as a fragmented ecosystem and slow technological innovation [4]. - The introduction of "Lingqu OS" is expected to integrate existing resources and promote ecosystem fusion, enabling developers to create applications on a unified platform [4]. Group 3: Future Prospects - The open-source plan for "Zhiyuan Lingqu OS" will begin in the fourth quarter of this year, aiming to drive innovation and transformation in the field of embodied intelligence [4]. - The initiative is anticipated to accelerate the large-scale application and popularization of embodied intelligent robots across various sectors, similar to the impact of Windows in the PC era and Hongmeng in the mobile internet era [4].

Core Viewpoint - Qiming Venture Partners is recognized as one of the earliest and most comprehensive investment institutions in the AI sector in China, having invested in over 100 AI projects, covering the entire AI industry chain and promoting the rise of several benchmark enterprises in the field [2]. Group 1: AI Models - In the next 12-24 months, a context window of 2 million tokens will become standard for top AI models, with more refined and intelligent context engineering driving the development of AI models and applications [4]. - A universal video model is expected to emerge within 12-24 months, capable of handling generation, reasoning, and task understanding in video modalities, thus innovating video content generation and interaction [6]. Group 2: AI Agents - In the next 12-24 months, the form of AI agents will transition from "tool assistance" to "task undertaking," with the first true "AI employees" entering enterprises, participating widely in core processes such as customer service, sales, operations, and R&D, thus shifting from cost tools to value creation [8]. - Multi-modal agents will increasingly become practical, integrating visual, auditory, and sensor inputs to perform complex reasoning, tool invocation, and task execution, achieving breakthroughs in industries such as healthcare, finance, and law [9]. Group 3: AI Infrastructure - In the AI chip sector, more "nationally established" and "nationally produced" GPUs will begin mass delivery, while innovative new-generation AI cloud chips focusing on 3D DRAM stacking and integrated computing will emerge in the market [11]. - In the next 12-24 months, token consumption will increase by 1 to 2 orders of magnitude, with cluster inference optimization, terminal inference optimization, and soft-hard collaborative inference optimization becoming core technologies for reducing token costs on the AI infrastructure side [12]. Group 4: AI Applications - The paradigm shift in AI interaction will accelerate in the next two years, driven by a decrease in user reliance on mobile screens and the rising importance of natural interaction methods like voice, leading to the birth of AI-native super applications [14]. - The potential for AI applications in vertical scenarios is immense, with more startups leveraging industry insights to deeply engage in niche areas and rapidly achieve product-market fit, adopting a "Go Narrow and Deep" strategy to differentiate from larger companies [15]. - The AI BPO (Business Process Outsourcing) model is expected to achieve commercial breakthroughs in the next 12-24 months, transitioning from "delivery tools" to "delivery results," and expanding rapidly in standardized industries such as finance, customer service, marketing, and e-commerce through a "pay-per-result" approach [15]. Group 5: Embodied Intelligence - Embodied intelligent robots will first achieve large-scale deployment in scenarios such as picking, transporting, and assembling, accumulating a wealth of first-person perspective data and tactile operation data, thereby constructing a closed-loop flywheel of "model - ontology - scene data," which will drive model capability iteration and ultimately promote the large-scale landing of general-purpose robots [17].

引领具身智能的规模化落地和实际应用。 本文为IPO早知道原创 作者｜ Stone Jin 微信公众号｜ipozaozhidao 演示。 事实上，这也是WAIC 2025现场极其少见的：机器人通过模型完成一整套长序列复杂操作，在开放 随机的环境里真正做到自主感知、决策与高精度操作。自变量将持续和产业上下游伙伴共建生态，引 领具身智能的规模化落地和实际应用。 本文由公众号IPO早知道（ID：ipozaozhidao）原创撰写，如需转载请联系C叔↓↓↓ 据IPO早知道消息，自变量机器人携基于自研端到端具身智能大模型的通用机器人亮相WAIC 2025 ——机器人"小白"与"小量"形成多机智能群，面对展会现场开放环境的动态变化，自主规划并完成 了一整套长序列复杂任务。分则各自完成家务整理、香囊制作，合则一起接力香囊补货。 其中，基于自变量机器人自研的通用具身大模型WALL-A，机器人"小量"在短短几天时间内就学会 自主制作香囊，不论是应对周遭复杂的声光环境，还是面向来回走动的人群，抗干扰能力极强。还会 根据观众的喜好进行个性化香囊制作，自主拾取对应颜色的香包和香材内胆。 而当向机器人发出简单的语音指令："小白，客厅有点 ...

标志着企业知识管理正式迈入 "智能体协同" 时代。 本文为IPO早知道原创 作者｜C叔 微信公众号｜ipozaozhidao 据IPO早知道消息，达观数据携全新AI产品与解决方案亮相WAIC 2025。其中，达观数据现场发布 的首款深度融合 Agent 能力的企业级知识库产品，标志着企业知识管理正式迈入 "智能体协同" 时 代，为金融、制造、能源、政务等多行业的知识管理升级开辟全新路径。 达观数据本次推出的 AI Agent办公智能体，通过四类专属Agent实现办公场景的智能化渗透，彻底 打破传统知识库"被动查询"的局限 ：1、审核Agent：化身"智能参谋"，辅助业务审核流程，自动 识别逻辑漏洞、查缺补漏，大幅提升合规性与准确性；2、问答类Agent：支持自然语言交互，快速 响应数据查询、生成报告/方案等文档，将人工检索效率提升数倍；3、填写Agent：聚焦规则明 确、重复机械的表单/报告填写任务，标准化输出结果，既降本又避免人为误差；4、归纳分析 Agent：从海量资料中自动提炼核心信息、归纳规律，让知识梳理从"耗时费力"变为"高效精准"。 值得注意的是， 这些Agent并非独立工具，而是深度融合到达观企 ...

Core Viewpoint - The article discusses the advancements in embodied intelligence by Zhiyuan Robotics, highlighting the establishment of a "flywheel system" that integrates data, models, entities, and scenarios to drive innovation across various industries [2][3]. Group 1: Embodied Intelligence Development - Zhiyuan Robotics has achieved a closed-loop development model by integrating robotic entities, motion intelligence, interaction intelligence, and operational intelligence, referred to as "one body, three intelligences" [3]. - The company has created the largest global dataset, AgiBot World, through its own data collection factory, aiming to address the data scarcity in embodied intelligence [3]. - The launch of the universal embodied base model, Qiyuan, allows for adaptability across different robotic entities, enhancing the intelligence and capabilities of robots [3]. Group 2: Application and Impact - Zhiyuan Robotics has successfully implemented its "robot + embodied model" technology in four key scenarios: industrial manufacturing, warehousing logistics, power inspection, and interactive guidance [4]. - The introduction of the Genie Envisioner platform marks a significant shift from passive execution to proactive "imagine-validate-act" capabilities for robots, enhancing their operational efficiency [6][9]. - The GE platform utilizes a multi-perspective video diffusion model, GE-Base, which is based on over 1 million video streams, enabling robots to perform tasks with high precision and robustness [6][8]. Group 3: Future Developments - The upcoming release of the next-generation robot body, G2, promises improvements in motion accuracy and scene adaptability, expanding the application boundaries of embodied intelligence in various environments [9].

Core Insights - The article highlights the core advantages and diverse application scenarios of the robotics industry showcased by Galaxy General at WAIC 2025, covering retail, industrial, logistics, and urban services [3]. Group 1: Robotics Capabilities - Galaxy General's exhibition featured a 1:1 replica of a real supermarket environment, where the Galbot operated continuously to demonstrate its capabilities [4]. - The Galbot's performance in complex environments is driven by the end-to-end embodied model, GroceryVLA, which allows for autonomous product recognition and stable grasping without the need for path planning [7]. - The model supports a unified grasping strategy across various product types, including soft and hard packaging, showcasing its versatility in retail applications [7]. Group 2: Industrial Applications - In the automotive parts sorting task, Galbot effectively handled challenges that traditional robots struggle with, such as identifying and grasping similar-looking parts and avoiding obstacles [8]. - Galbot demonstrated its ability to self-correct and adapt to disruptions, such as changes in the arrangement of parts, highlighting its autonomous decision-making capabilities [10]. - The robot's efficiency in box handling in industrial and logistics settings matched that of human workers, emphasizing its practical application value in complex environments [12]. Group 3: Urban Services - Galaxy General's self-developed model enabled a robotic dog to autonomously pick up litter and conduct inspections, showcasing its adaptability to dynamic environments [12].

Core Viewpoint - The article highlights the advancements and applications of AI² Robotics' AlphaBot series, showcasing its capabilities in various industrial and service sectors, emphasizing the importance of practical applications over mere technological demonstrations [2][4][7]. Group 1: Technological Innovations - The AlphaBot series features a hardware form and a foundational model called Alpha Brain, which enables multi-tasking and spatial awareness through advanced technologies [4]. - The GOVLA model allows for end-to-end closed-loop control, integrating multi-modal information for seamless perception and action, overcoming traditional robotic limitations [4][6]. - AI² Robotics has developed the RoboMamba model and the FiS-VLA model, which significantly enhance the generalization capabilities and response speed of robots in complex environments [6]. Group 2: Industrial Applications - AI² Robotics has partnered with Dongfeng Liuzhou Motor to implement AlphaBot in various manufacturing processes, marking a significant milestone for domestic models in automotive manufacturing [9]. - In the biotechnology sector, AlphaBot is being deployed in sterile environments for material handling and visual inspection, reducing contamination risks and adapting to changing processes [10]. - The AlphaBot has also entered the semiconductor manufacturing space, efficiently executing tasks at the production base of Geely Technology [10]. Group 3: Expansion into Public Services - AI² Robotics plans to introduce AlphaBot into major airports in first-tier cities, demonstrating its autonomous capabilities in complex public environments [11]. - The company's approach focuses on addressing real industry needs and continuously refining robot performance through practical applications [11]. Group 4: Vision for the Future - The founder and CEO of AI² Robotics envisions general-purpose intelligent robots becoming essential smart terminals in daily life, akin to smartphones and smart cars [12].

Core Viewpoint - The article emphasizes the role of domestic computing power in enabling innovative internet applications, highlighting the achievements of Suiyuan Technology in AI computing infrastructure and commercialization [2][5]. Group 1: AI Computing Infrastructure - Suiyuan Technology showcased its latest achievements in AI computing infrastructure at WAIC 2025, focusing on the theme "The Fire of Chips Spreads" [2]. - The company has established intelligent computing centers in various locations, including Qingyang, Wuxi, and Yichang, which support the development of domestic AI [5]. - By the end of 2024, Suiyuan Technology built the first 10,000-card inference cluster in Qingyang, providing robust support for the "East Data West Computing" initiative [5]. Group 2: Product Offerings - The exhibition featured the "Suiyuan® S60" AI inference card, which has been widely applied in various internet scenarios, including chatbots and online meeting summaries [2]. - The DeepSeek integrated machine series, set to launch in early 2025, supports domestic CPU platforms and offers scene optimization capabilities for efficient AI business deployment [3]. Group 3: Commercialization and Partnerships - Since 2020, Suiyuan Technology has collaborated with Tencent to explore the commercialization of domestic computing power across different business scenarios, delivering tens of thousands of cards for high-demand applications [5]. - The company has successfully deployed inference cards for the "AI dressing" feature of Meitu's beauty camera, ensuring stable performance during peak usage [6]. - Suiyuan Technology is actively converting domestic AI computing power into actual commercial value through deep collaboration with various clients [7]. Group 4: Industry Insights - The rapid development of large models is transforming the industry ecosystem, driving the systematic and clustered development of computing infrastructure [9]. - Suiyuan Technology aims to provide inclusive, efficient, and reliable AI infrastructure solutions, fostering an open-source domestic AI ecosystem in collaboration with industry partners [9].

Core Viewpoint - Lingchu Intelligent has made significant advancements in the field of embodied intelligence, showcasing its comprehensive technology chain from data, model algorithms, hardware to application scenarios at WAIC 2025 [2][13]. Group 1: Product Demonstrations - The Mahjong Robot demonstrated the ability to engage in a 30-minute continuous game with human players, showcasing its strategic thinking and decision-making capabilities in a complex environment [2][3]. - The Autonomous Packing Task showcased the robot's ability to understand natural language commands and autonomously execute tasks such as analyzing item arrangement and completing packaging without manual intervention [4]. - The Delivery Robot addressed the "last mile" delivery challenge by accurately identifying and handling various delivery items, demonstrating its flexibility and automation capabilities [6][7]. Group 2: Core Technologies - The 21-degree-of-freedom dexterous hand was highlighted for its exceptional operational dexterity and efficiency, capable of high-precision control and tactile feedback, meeting industrial demands for fine operations [9]. - The Exoskeleton Remote Operation solution was introduced, allowing users to control robotic hands with high precision, contributing to a high-quality data collection platform for reinforcement learning [11]. Group 3: Industry Positioning - Lingchu Intelligent has established a leading position in the L3 long-range dexterous operation sector, marking a significant transition from simple action execution to cognitive decision-making and complex task handling [13]. - The company has built a robust technology ecosystem consisting of four pillars: a data pyramid combining simulation and real data, the VLA model based on the CoAT framework, a cost-effective hardware system, and comprehensive validation across various scenarios [13].