The 2025 Central Digital Economy Industry Development Conference (referred to as "Central Digital Expo") grandly opened in Zhengzhou, Henan. This industry event, centered on the theme "Gathering Digital Intelligence Forces to Boost Central China's Rise," brought together core forces from government, enterprise, and academic circles to focus on key issues such as data element value release and industrial digital transformation. As a leader in enterprise-level Data&AI technology, KeenData was invited to attend this conference. Guo Zhenqiang, Co-founder and Vice President of KeenData delivered an in-depth presentation at the conference on data element value release, digital transformation capability building for large organizations, and Data&AI infrastructure technology implementation practices, demonstrating the company's technical strength in the data infrastructure field and providing implementable solutions for digital transformation across various industries.
Guo Zhenqiang, Co-founder and Vice President of KeenData
Industry Demands Drive Upgrades: Data&AI Infrastructure Enters Critical Implementation Phase
Currently, the digital economy has comprehensively shifted from the "technology exploration" stage to the "practical implementation" stage, with the value release of data elements becoming the core demand for industrial upgrading. With the deep penetration of AI technology, unstructured data accounts for over 80%, and the processing demands for multimodal data such as audio-video, IoT signals, and documents have surged. Traditional databases from the IT era and transitional data platforms from the cloud computing era can no longer meet complex requirements such as "integrated model training and inference" and "full-scenario data intelligence."
National policies have injected strong momentum into practical implementation. The "Trusted Data Space Development Action Plan (2024-2028)" clearly requires the implementation of over 100 trusted data spaces by 2028, covering five major scenarios including enterprises, industries, and cities. The "Data Elements ×" Three-Year Action Plan emphasizes scenario-driven data circulation and deepens the application of privacy computing, blockchain, and other technologies. Against this backdrop, building AI-native Data&AI integrated infrastructure is no longer a "choice" but a "necessity" for enterprises to break through transformation bottlenecks and for regions to achieve digital economy breakthroughs.
However, current large models, big data, and AI applications generally face core pain points such as low value conversion of unstructured data, large model inference performance failing to meet application demands, and AI applications not yet effectively converted into productivity improvements. To address these industry challenges, KeenData, leveraging its technical accumulation and practical experience in the Data&AI field, has conducted a series of forward-looking explorations and practices, focusing on building an efficient and intelligent enterprise-level Data&AI integrated platform KeenData Lakehouse to help enterprises break through digital transformation bottlenecks, truly achieve cost reduction and efficiency improvement with business innovation, and drive deep transformation of data element value into actual productivity.
Core Interpretation: Building Data&AI Integrated Data Infrastructure
KeenData Lakehouse adopts an AI-Native intelligent-driven architecture to achieve Data&AI engineering integration capabilities. The platform targets large organizations for systematic data and AI implementation, providing infrastructure products covering the full chain from data integration, offline and real-time development, multimodal computing, data governance, dataset management, AI model building, integrated training and inference to Agent development. The platform breaks through traditional architectures that separate data and AI, with self-developed AI-in-Lakehouse technology unifying lakehouse engines, OLAP data governance, and AI technology to form a streamlined and efficient All-in-One technical solution. The self-developed multimodal computing engine completes data cleaning to result analysis in a single pipeline, multiplying GPU inference throughput, combined with KMI inference acceleration, model quantization, and unified catalog (Unity Catalog) to achieve cross-modal intelligent governance.
Data&AI Integration
The platform achieves seamless integration of data full-lifecycle processing with AI development processes through deep integration of data and AI, forming a closed-loop capability of data processing-AI development-application implementation.
Its core characteristics are reflected in three aspects:
Multimodal Data Processing: Supports text/image/audio-video fusion governance;
Agent Intelligent Architecture: Achieves perception-cognition-action-evolution closed loop;
Data&AI Integration: Data&AI native fusion provides All-in-One architecture capabilities, eliminating the separation between Data and AI architectures.
(Figure: Data&AI Integration)
AI-Native
Unlike traditional platforms' loosely coupled external AI mode, KeenData's Data&AI integrated platform takes AI-Native as its core design philosophy, deeply embedding intelligent capabilities into the system's genes, building an intelligent data foundation with autonomous evolution capabilities—its technical architecture and core capabilities are all developed around the bidirectional drive of AI efficiently processing data and data intelligently supporting AI, covering three core capabilities: MaaS self-inference, Agent self-iteration, and data full-lifecycle intelligence.
Addressing the pain points of traditional compute-storage integrated architectures such as low resource utilization and high expansion costs, the platform adopts a compute-storage separation architecture, with data uniformly stored in high-performance unified storage and computing resources elastically scalable on demand, not only reducing storage costs by over 30% but also allowing AI training, inference, and other computing tasks to flexibly call resources, completely solving the resource contention problem of large tasks crowding out small tasks, laying a solid foundation for the implementation of intelligent closed-loop capabilities.
(Figure: AI-Native Full-Chain Capability Closed Loop)
From Projects to Value: KeenData Lakehouse Multi-Scenario Practical Answers
Municipal Data Bureau Data Infrastructure Project: Based on KeenData's Data&AI integrated platform, non-algorithm teams can prepare data through the corpus processing layer, use the intelligent support layer to complete model training, fine-tuning, and deployment with zero code, then call APIs or build agents to quickly convert large models into commercial products. At the same time, it connects the full chain from multimodal data to industry agents, covering the full lifecycle of "data→model→application," supporting rapid construction of data products for small-incision scenarios. Through standardized SDK and plugin interfaces, it opens third-party corpus processing tool access, achieving "plug-and-play," accelerating the implementation efficiency of large models in urban business scenarios and promoting AI technology to truly serve urban governance and industrial upgrading.
Municipal Digital Government 2.0 Project: Based on the trusted data space built on KeenData's Data&AI integrated platform, the new smart city big data infrastructure and trusted data space achieve comprehensive digital, intelligent, and refined management and services in government affairs, people's livelihood, industry, and other fields, while ensuring that data resources are deeply mined and rapidly applied, thereby driving the rapid development of cities and even city clusters, building the first government-side intensive data infrastructure common support platform, exploring effective supply of government public data to social enterprises.
Central State-Owned Enterprise Data Intelligence Foundation Project: Relying on KeenData's Data&AI integrated platform, a unified data center and governance system was built, completing efficient storage and computing of newly added big data, and further combining with business scenarios to provide hundreds of service supports for planning, engineering decision-making, and engineering integrated platforms. AI-driven full-volume business and research data management sharing accelerates the digital transformation of data into resources and assets, improves operational efficiency, achieves integrated chain operations, and is an important milestone for the group's digital intelligence operations entering a new stage of efficient collaboration.
Facing the opportunities and challenges of the data intelligence era, KeenData focuses on deep integration innovation of Data&AI, committed to building a solid digital intelligence transformation foundation for large organizations and enterprises. To date, KeenData has provided Data&AI infrastructure construction for nearly 200 large domestic and international organizations across 20+ industries, including Fortune 500 companies such as Sinopec, China Unicom, China Telecom, CITIC Bank, AEON Group, FAW Group, and Sany Heavy Industry, based on the KeenData Lakehouse platform. Looking to the future, KeenData will continue to drive industrial transformation through technological innovation with KeenData Lakehouse, helping large organizations and enterprises release "AI new quality productivity," seize competitive advantages in the digital era, and jointly promote the entire industry toward a new journey of intelligent development.
