SenseCore’s CGO: RaaS Enables Large Models to Truly Enter Industry

Recently, the “Digital Intelligence Transformation · AI New Journey” 2026 CIO Times New Year Forum, jointly organized by CIO Times and the New Infrastructure Innovation Research Institute (新基建创新研究院), was held in Beijing.

Yang Song, Chief Growth Officer (CGO) of the SenseCore Business Group, delivered a keynote speech titled “Practices and Reflections on Result as a Service (RaaS) for Large Models” at the main forum. His presentation systematically explored the pathways, business logic, and methodologies for large models in industry, drawing significant attention from participating CIOs and industry representatives.

Yang Song emphasized that the development of large models has entered a critical stage: “Ultimately, large models must return to delivering industrial value and results. Standardized models and platforms alone cannot fully address the real business challenges of customers.” He stressed that for AI to truly integrate into industry, it must shift from “capability demonstration” to “Results as a Service (RaaS).”

Addressing Business Pain Points: The First Principles of Large Model Implementation

“Whether in the early Computer Vision (CV) era or in today’s large model era, clients have always judged with a single standard—whether there are clear, tangible benefits,” said Yang Song. Against this backdrop, SenseCore has progressively defined its approach to “Results as a Service (RaaS),” delivering business outcomes as the ultimate output rather than the model or algorithm itself.

He reiterated that the starting point for large model implementation lies not in the model itself, but in understanding the business.

“Only by deeply understanding clients’ production processes, decision-making logic, and key metrics can data truly be transformed into knowledge and embedded into business systems through intelligent agents,” noted Yang Song. In real industrial environments, standardized solutions often fail to address complex and dynamic business scenarios. Expert teams must therefore engage closely with customers to identify challenges and design tailored pathways.

ROI Is Achieved Through Practice, Not Calculation

Addressing the widespread concern over return on investment (ROI), Yang Song offered a straightforward perspective: “The ROI of AI applications is not calculated—it is achieved through practice.”

He explained that in complex industrial settings, technological evolution and business changes often occur simultaneously. Static, one-time ROI assessments cannot reflect true value. SenseTime has therefore developed a practical mechanism centered on Proof of Concept (POC) validation, upfront evaluation, and dynamic adjustments, treating ROI management as a continuous process jointly undertaken with customers.

This results-oriented delivery model essentially functions as a proactive quality control and risk management mechanism, ensuring AI projects are effectively integrated into production systems and generate stable business returns.

Building a Replicable RaaS Capability System

Drawing on extensive industry experience, SenseCore has established a comprehensive capability system encompassing AI infrastructure, large model platforms, data and knowledge governance, and industry expert services.

Yang Song noted that this system both supports successful delivery of individual projects and enables replication and scalable implementation of RaaS across industries and varying levels of complexity.

In practice, SenseCore has already developed benchmark cases in several domains.

For example, in collaboration with the China Railway First Survey and Design Institute Group, SenseCore built the industry’s first multimodal large model application platform, integrating expertise across 28 industries. The platform addresses long-standing challenges, such as knowledge transfer difficulties, poor integration, and low application rates.

In urban governance, SenseCore supported the Shanghai Municipal Bureau of Planning and Natural Resources in developing the nation’s first foundational large model in the planning and resource sector—the “Yunyu Xingkong Large Model” (雲宇星空大模型). This technical framework integrates spatiotemporal understanding and generation, knowledge integration and retrieval, and intelligent model scheduling, serving as an “AI partner” for industry professionals.

At the forum, SenseCore was recognized with the “Annual CIO Most Trusted Digital Partner Award” for its systematic capabilities and achievements in AI industrialization.

Yang Song concluded that business scenarios determine the value of data and models, and that results orientation is the core of AI implementation. Only AI projects built on clear business definitions and quantifiable analyses can achieve long-term stability and sustainable success.