Customer Service Automation
Shanghai Pudong Development Bank needed a scalable, contactless way to deliver personalized wealth-management guidance and everyday banking service to a very large retail customer base without forcing customers to wait for human staff.
Reported impact
The bank introduced a reusable digital-human service channel that scaled personalized advisory and service interactions, diverted most routine traffic from human agents and created a foundation for broader future-bank automation.
Shanghai Pudong Development Bank and Baidu AI Cloud built Xiaopu as a digital employee that spans the bank app, online banking and service terminals. The implementation starts with a 3D twin of a real wealth manager, using high-precision body capture, voice cloning and text-to-speech so the avatar can speak naturally, while facial-expression recognition and full-duplex speech let customers interrupt and continue multi-turn conversations hands-free. Behind the avatar, Baidu AI Cloud's AI middle-platform and knowledge middle-platform combine user profiles, risk preferences, product knowledge, recommendation logic and accumulated dialogue records into a one-to-one advisory workflow. The system continuously learns from conversation data, improves customer-intent understanding and recommendations over time, and extends the same stack into the bank's broader future-bank concept so digital staff can front sales, service and advisory scenarios across multiple devices.
Customer voice and profile inputs from the mobile app, online banking and terminal channels flow into Xiaopu's digital-human interface, where acoustic and semantic analysis call the AI middle platform, user-profile and recommendation services, and the knowledge middle platform returns tailored answers and product suggestions.
As the industry's first 3D twin digital human, Xiaopu restores a real wealth manager's appearance, voice, expressions and gestures (original: 作为行业首个 3D 孪生数字人，小浦实际上是真实地还原了一位理财经理的相貌、声音、表情、手势动作).
If Xiaopu serves 10,000 customers a day, the bank can save about 3,333 labor hours per day and nearly 10,000 person-days per month (original: 若小浦服务一位客户的时长为 20 分钟且每天服务 1 万名客户，可每天节约人工 3333 小时……每月节省人力近 1 万人).
The bank's first digital employee, Xiaopu, has already appeared in the app, online banking and various service terminals (original: 目前，浦发银行和百度智能云联合打造的银行业首个数字人员工小浦已经出现在 App、网银和各类服务终端上).
Xiaopu gives users the phone's only full-duplex visual interaction experience and supports multi-turn dialogue after the wake phrase 'Xiaopu, Xiaopu' (original: 小浦提供给用户的是手机端唯一的全双工可视化交互新体验……客户用“小浦小浦”即可唤醒数字人，实现多轮对话).
Xiaopu's voice customer service already accounts for more than 80% of the bank's total customer-service traffic (original: 小浦在全行客服渠道上线，其话务量已占全部话务量八成以上).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Baidu AI Cloud Sample Jun 30, 2025 Industrial Bank Co., Ltd. (兴业银行股份有限公司) Industrial Bank wanted its digital-service channels to deliver better customer experience, broader self-service coverage, and more precise operations as customers moved online, but legacy service models could not scale… Average daily handled volume reached nearly 100,000 calls. Baidu AI Cloud Intelligent Customer Service · Intelligent Speech · Smart Outbound Calling Baidu AI Cloud Sample Jun 27, 2025 CITIC Financial Holdings CITIC Financial Holdings needed to make its wealth-management platform more intelligent and scenario-driven so customer-service and wealth-adviser interactions could scale across a very large client base. End-to-end question-answering accuracy exceeded 85%. Baidu AI Cloud Keyue · Knowledge Expansion Scheme · Knowledge Flywheel Baidu AI Cloud Sample May 14, 2025 China Life Property & Casualty Insurance Co., Ltd. (中国人寿财险) China Life P&C needed a scalable way to collect, analyze, and act on large volumes of customer NPS survey data across policy, claims, and value-added service touchpoints, especially the unstructured feedback that… Customer sentiment recognition accuracy exceeded 95%. BML Full-Featured AI Development Platform · PaddlePaddle · Sugar BI 