转自：前瞻产业研究院 行业主要上市公司： 科大讯飞(002230)、第四范式(06682)、拓尔思(300229)、用友网络(600588)、云从科技 (688327)、出门问问(02438)、迈富时(02556)等 本文核心观点：中国AI代理行业已形成"SaaS-MaaS-RaaS"三足鼎立的商业模式，在技术、政策、生态三重 驱动下，正通过区域差异化竞争加速万亿级市场的商业化落地。 SaaS主导、MaaS狂奔、RaaS深耕的AI代理变现图谱 我国中国AI Agent行业商业模式可按服务形态、部署方式及应用场景分为三大类，其中SaaS模式占据主导地 位，主要得益于企业对标准化智能工具的需求、MaaS模式增速最快，反映出模型即服务的商业化加速、 RaaS模式在制造业和金融领域渗透率提升显著。 | 商业模式 | 市场占比 | 代表企业 | 核心应用场景(1 | | --- | --- | --- | --- | | SaaS 模式e | 30% | 百度智能云、Salesforcee | 智能客服、办公自动化(2 | | MaaS 模式(2 | 15% <= | 商汤科技、科大讯飞(2 | 模型训练、推理服务(2 ...

Core Insights - The customer service team at Pinduoduo plays a crucial role in maintaining user experience and resolving transaction disputes, but they face significant pressure, especially during peak promotional periods [1][3][5] Group 1: Sources of Pressure on Customer Service - The volume of inquiries surges geometrically during promotions and new product launches, overwhelming the customer service team [3] - A large proportion of customer inquiries consist of repetitive, standardized questions, leading to inefficiencies and potential burnout among staff [4] - Customer service representatives often bear the brunt of negative emotions from dissatisfied users, requiring strong emotional management skills [5] - The rapid changes in platform rules and product information necessitate continuous learning, adding to the workload and stress of customer service personnel [6] Group 2: Role of Intelligent Customer Service Agents - Intelligent Customer Service Agents (AI) are emerging as a key solution to alleviate the pressures faced by human customer service representatives [6] - These AI agents can operate 24/7, effectively handling a large volume of simple inquiries, especially during peak times, allowing human agents to focus on more complex issues [7] - AI agents serve as intelligent assistants, providing standardized responses to frequently asked questions, thus freeing human agents from repetitive tasks [9] - Advanced AI agents possess emotional analysis capabilities, enabling them to identify and manage user emotions, which helps mitigate the emotional burden on human agents [9] Group 3: Human-Machine Collaboration - The goal of intelligent customer service agents is not to replace human agents but to work collaboratively, enhancing overall service quality and efficiency [8] - By filtering out low-value inquiries and providing real-time support, AI agents enable human representatives to handle more sensitive and complex issues with greater confidence [9] - The integration of AI in customer service represents a future direction for e-commerce platforms, improving user experience and operational efficiency [8][9]

Core Viewpoint - The article emphasizes that companies should focus on integrating AI Agents with business scenarios to create value rather than blindly pursuing technological iterations [2] Group 1: AI Agent Development Stages - The development of intelligent customer service can be divided into three stages: 1. **Traffic Interception**: The primary goal is to answer user questions without focusing on service quality [3] 2. **Service Level Improvement**: Enhancing the service level to that of a business expert through AI technology [3] 3. **User Experience Companion**: Evolving into a comprehensive shopping assistant that provides personalized support [3] Group 2: Deployment Efficiency - The introduction of generative AI has significantly lowered the deployment threshold for intelligent customer service, reducing setup time from about one week to just a few hours [4] - Currently, 90% of JD.com's self-operated customer service has adopted AI models, retaining only 10% of human agents [4] Group 3: Value Creation in Customer Service - The application of large models in intelligent customer service is not revolutionary but effectively reduces costs and increases efficiency [5] - Key factors for rapid application include: 1. **User and Scenario**: The vast number of user applications in intelligent customer service creates significant value [5] 2. **Data Availability**: The large volume of structured interaction data supports high-quality model training [5] 3. **Revenue Model**: The clear evaluation of ROI from replacing human labor with AI [5] Group 4: Specific Use Cases - Intelligent customer service has shown effectiveness in various scenarios, such as refunds and reshipments, with significant reductions in processing time and labor costs [6][7] - For example, the implementation of intelligent agents in refund processes has reduced processing time by 60% and decreased the workload of human agents by 60% [7] Group 5: Broader Applications - Beyond e-commerce, intelligent agents are also being utilized in government services, such as the 12345 hotline, improving response times and operational efficiency [8][9] Group 6: Current Limitations and Future Potential - Despite the advancements, intelligent customer service is still in the "L2+" stage, requiring human intervention for complex issues [10] - The future of intelligent customer service lies in creating a symbiotic relationship between digital employees and human experts, with a focus on integrating SaaS and Agent models [11]

Investment Rating - The report does not explicitly state an investment rating for the industry or the specific company being analyzed [1]. Core Insights - The company focuses on financial technology solutions and data center solutions, establishing partnerships with leading AI companies to maintain technological leadership and explore AI applications across various verticals [18]. - The company has developed multiple AI Agent products, including intelligent customer service and credit agents, and has launched an Agent development platform [18][21]. - The generative business process AI agent is a key innovation that integrates generative AI with business process management, aimed at enhancing operational efficiency and decision-making in banks [19][20]. Summary by Sections 1. Basic Research Information - The research was conducted on March 7, 2025, focusing on the internet services industry, specifically the company Zhongke Jincai [13]. 2. Detailed Research Institutions - Various institutions participated in the research, including Guosen Securities, Minghe Investment, and Changxin Fund, among others [14][15]. 3. Research Institution Proportions - The report includes a breakdown of the participating institutions, but specific proportions are not detailed [16]. 4. Main Content Information - The company is a leading service provider in the banking sector, leveraging its experience to develop generative business process AI agents that can optimize resource allocation and enhance operational flexibility [20][21]. - The generative business process AI agent can significantly reduce decision-making and product development cycles by automatically generating solutions based on real-time data [22]. - The company has a comprehensive model selection and evaluation system, ensuring the integration of the latest models into its solutions [21].