Customer Service Automation · Enterprise Knowledge Search
Chaoxing Zhizhen Group needed a way to connect isolated campus systems and fragmented knowledge resources, deliver more personalized services to teachers and students, and turn large-model technology into production-ready applications for teaching, campus management, and research workflows.
Reported impact
The AI capability center raised large-model output accuracy to nearly 90%, expanded campuses to more than 40 AI tools, drove high adoption of staff and student productivity tools, and cut agent-development effort and delivery time so schools can launch new smart-campus experiences faster and at lower operating cost.
Chaoxing Zhizhen Group worked with Alibaba Cloud to build a university AI capability center on Alibaba Cloud Bailian that connects data governance, knowledge management, agent delivery, and applications. The architecture is organized as end user, business scenario, and capability middle platform: campus systems, teaching resources, and management content are ingested into a centralized layer, where Bailian's multi-source data-governance tooling creates a dual-track knowledge system with institution-specific repositories plus a public shared library. On top of that foundation, the group packages AI functions into an AI agent platform, AI knowledge-base platform, and AI tool platform so universities can assemble multimodal Q&A assistants, recommendations, and business workflows without writing low-level model code. Qwen models and Bailian APIs provide the inference layer, while natural-language instructions and visual configuration shorten agent creation and rollout.
Campus information systems, teaching resources, and management content feed the AI capability middle platform, where Bailian's data-governance functions organize them into dedicated and public knowledge bases. Qwen-powered agents and tools then retrieve from those stores and return multimodal Q&A, recommendations, and workflow outputs through campus-facing applications.
Bailian's multi-source data-governance capabilities were used to create a dual-track knowledge system made of institution-specific repositories plus a public shared library. (original: 借助阿里云百炼的多源数据治理能力，打造超星指针“单位专属库+公共共享库”双轨知识体系。)
Chaoxing described isolated systems, fragmented knowledge, limited personalization, and weak AI-to-business integration as the main barriers in smart-campus transformation. (original: 系统数据割裂，形成信息孤岛... 服务模式单一，个性化能力不足... AI 融合乏力，技术落地门槛较高。)
The project lifted output accuracy to nearly 90%, provided more than 40 AI tools, raised PPT assistant usage to 80%, improved development efficiency 85%, shortened average development to 1 hour, and reduced labor cost 70%. (original: 借助 RAG 技术，大模型输出准确率提升至近 90%... AI 能力平台提供近 40+ 个 AI 工具... PPT 辅助工具使用率攀升至 80%... 开发效率提升 85%，平均开发周期缩短至 1 小时... 人力成本降低 70%。)
The solution packages complex AI technology into an AI agent platform, AI knowledge-base platform, and AI tool platform so business teams can use visual configuration instead of low-level engineering. (original: 基于阿里云百炼 API 服务，超星指针集团将复杂 AI 技术封装为 AI 智能体平台、AI 知识库平台、AI 工具平台三大核心模块。)
To solve those issues, the group and Alibaba Cloud built an education large-model capability center on Bailian that covers the full chain from model scheduling to application scenarios. (original: 为了解决上述挑战，超星指针集团选择与阿里云合作，基于大模型服务平台百炼，共同构建了面向高校的 AI 能力中心，实现了从模型调度到应用场景的全链路赋能。)
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.
“超星成立于1993年，现有员工5000余人，在全国31个省份均设置有分公司及办事处，在北京、上海、南京、成都、武汉、深圳、福州、济南等地建立10个研发中心。”
Related
Databricks Sample Mar 20, 2026 Baylor University Enrollment Management Baylor's Enrollment Management division serves 19,000 students and 50,000 annual applications, with a small contact-center team handling 100–300 calls per day about financial aid and student accounts. Dedicated QA could… 100% call review coverage (up from ~5%) Agent Bricks / Mosaic AI · Databricks Genie · Databricks Marketplace 
Microsoft Sample Dec 6, 2024 SEDUC São Paulo SEDUC has an overwhelming amount of student essays. They required personalized content creation. They wanted to decentralize access to data. And if you have about 40 students in each of these classes, we are talking about 240 students. Azure OpenAI Service · Dynamics 365 · Power Platform 
Bain Sample Archived · Nov 23, 2024 Multiversity S.p.A. Multiversity wanted to reinforce its leadership in educational innovation, but it needed to prove that generative AI could answer students accurately and improve staff efficiency before scaling the technology across its… 90% faster completion of administrative tasks AI University Chatbot · Open AI technology · custom tool 
