What kind of innovative sparks will fly when AI meets Dongcheng Group, the top 5 chain hotel company?

On July 28, at the "AI Doers" DingTalk session of the 2025 World Artificial Intelligence Conference, Liu Yan, Vice President and Chief Technology Officer of the Qingmao Hui Business Group at Dongcheng Group, delivered a speech titled "Dongcheng's Exploration in AI Innovation: AI Empowering Innovation in the Hotel Industry," sharing the group's innovative practices and significant achievements in the field of artificial intelligence.

Deepening AI Applications to Address Industry Challenges

As a top-five player in the hotel industry with 3,500 properties, 250,000 rooms, and more than 100 million members, Dongcheng Group is leveraging DingTalk's intelligent foundation to advance AI applications in depth, addressing core challenges of cost reduction and efficiency improvement brought about by economic downturn, consumption downgrade, and intense industry competition.

Focusing on Three Key Areas to Enhance Operational Efficiency

Facing pressure from per-room construction costs, supply chains, and operating expenses, as well as the need to improve operational metrics such as staff-to-store ratio and staff-to-room ratio, Dongcheng has identified digitalization and intelligence as a critical path. The company is focusing on three main areas: internal efficiency improvement, AI innovation, and low-code applications. By promoting AI programming, development efficiency has increased by about 12%. In the low-code arena, leveraging the DingTalk platform enables efficient integration of non-core systems and task supervision, boosting efficiency by 60% compared to full-code development—exceeding expectations.

Building an "AI Startup Lab" to Drive Innovation Implementation

Dongcheng has innovatively established an "AI Startup Lab," creating four major support systems: organizational safeguards, a technical foundation, talent development, and operational promotion. The lab collaborates with business teams and DingTalk AI experts to accelerate the rapid implementation and scenario incubation of AI technologies within the company. Currently, 10 business units and 17 business pioneers are involved, proposing two to three dozen business ideas, with about 15 already entering the implementation phase.

A Tiered Technical Architecture Ensures Safe and Efficient AI Operations

In terms of technical architecture, Dongcheng has built a tiered system: the data layer uses an independent data platform to ensure the security of AI calls; the knowledge layer is unified and integrated into the DingTalk platform, where it undergoes cleansing and control; the development layer combines the DingTalk productivity platform, a self-built AI development platform, and traditional machine learning platforms to meet application needs of varying complexity.

Two Core Applications Enhance C-End Experience and Store Operations

At the conference, Liu Yan highlighted two core AI applications: "AI Qingmao Elf," designed for guests, and "AI Super Store Manager," which empowers store operations. "AI Qingmao Elf" features four layers of capabilities: instantly answering simple questions; using a knowledge base to handle general inquiries; providing personalized responses based on user profiles; and connecting to a ticketing system to handle requests such as room service and cleaning. It also takes on pre-sales screening and customer follow-up marketing tasks, effectively addressing slow human service response times, inconsistent standards, and high labor costs.

"AI Super Store Manager" reshapes the store manager's workflow: first, it automatically extracts key operational data (such as daily dashboards and OTA health diagnostics), simplifying information presentation; second, it enables intelligent business processing, such as automating long-term pricing adjustments for 8–60 days in advance and pushing 1–7 day pricing recommendations for decision-making; third, it uses DingTalk cards or phone alerts to provide real-time warnings about competitor price changes or traffic fluctuations, ensuring revenue maximization. This frees store managers from tedious tasks and enables efficient, intelligent decision-making.