The 2025 Central Digital Economy Industry Development Conference, also known as the Central Digital Expo, opened in Zhengzhou, Henan. With the theme “Gathering Digital Intelligence to Boost the Rise of Central China”, the event brought together key representatives from government, enterprises and academia to focus on essential topics such as data element value realization and industrial digital transformation. As an enterprise-level leader in Data and AI technologies, KeenData was invited to participate. At the conference, Guo Zhenqiang, Co-founder and Vice President of KeenData, delivered an in-depth presentation on data-element value realization, capability building for digital transformation in large organizations and practical implementation of Data and AI infrastructure technologies. His session showcased the company’s technological strength in data infrastructure and provided actionable solutions to support digital transformation across industries.

Today, the digital economy has fully shifted from the stage of technological exploration to the stage of practical implementation, and the release of data-element value has become a central requirement for industrial upgrading. With the deep penetration of AI technologies, the proportion of unstructured data has surpassed 80 percent. The demand for processing multimodal data such as audio and video, Internet of Things signals and documents is rising rapidly. Databases from the traditional IT era and transitional data platforms from the cloud computing era can no longer meet complex requirements such as integrated model training and inference or full-scenario data intelligence.

National policies are also providing strong momentum for real-world implementation. The Trusted Data Space Development Action Plan (2024–2028) sets the goal of establishing more than one hundred trusted data spaces by 2028 across enterprise, industry and urban scenarios. The Data Element Plus Three Year Action Plan emphasizes scenario driven data circulation and calls for deeper application of technologies such as privacy preserving computation and blockchain. Under this direction, building AI native Data and AI integrated infrastructure is no longer an optional choice. It has become the essential path for enterprises seeking to overcome transformation bottlenecks and for regions aiming to achieve breakthroughs in the digital economy.

However, current large models, big data systems and AI applications are facing several fundamental challenges. These include low conversion efficiency of unstructured data into usable value, insufficient inference performance of large models to meet real application requirements and the inability of many AI applications to translate effectively into productivity gains. In response to these challenges, KeenData has leveraged its accumulated technological expertise and extensive practical experience in the Data and AI domain to conduct forward looking research and implementation. The company focuses on building the highly efficient and intelligent enterprise level Data and AI integrated platform, KeenData Lakehouse, which helps enterprises overcome digital transformation bottlenecks, achieve real cost reduction and efficiency improvement and drive business innovation. This enables the value of data elements to be deeply transformed into actual productive power.

KeenData Lakehouse adopts an AI Native intelligent driven architecture, which enables unified Data and AI engineering capabilities. Designed for large organizations to operationalize data and AI systems in a systematic manner, the platform provides foundational infrastructure products which cover data integration, batch and streaming development, multimodal computation, data governance, dataset management, AI model building and the full chain closed loop from training and inference to agent development. The platform breaks through traditional architectures in which data and AI are separated. Its self developed AI in Lakehouse technology unifies the lakehouse engine, OLAP data governance and AI technologies, which forms a streamlined and highly efficient All in One technical solution. The self developed multimodal computation engine completes the entire process from data cleaning to analytical output in a single pipeline, which increases GPU inference throughput by several times. Together with KMI inference acceleration, model quantization and a unified catalog capability similar to Unity Catalog, it achieves cross modal intelligent governance.

The platform achieves deep fusion between data and AI, which connects full lifecycle data processing with the AI development workflow and which forms a closed loop that spans data processing, AI development and application deployment. Its core characteristics are reflected in three dimensions:

• Multimodal data processing, which supports unified governance for text, images, audio and video.
• Agent centric intelligent architecture, which enables a complete loop of perception, cognition, action and evolution.
• Data and AI integration, which provides a native All in One architecture that eliminates fragmentation between data systems and AI systems.

Distinct from traditional platforms which adopt loosely coupled external AI components, the Data and AI integrated platform from KeenData adopts AI Native principles at its core, which embeds intelligent capabilities deeply into the system foundation and which builds an intelligent data infrastructure capable of autonomous evolution. Its technical architecture and core capabilities are built around a dual driven mechanism in which AI efficiently processes data and data intelligently supports AI. This covers three major capabilities: MaaS autonomous inference, agent self iteration and intelligent enablement across the entire data lifecycle.

To address the pain points of traditional storage compute integrated architectures, which lead to low resource utilization and high expansion costs, the platform adopts a storage compute separation architecture. Data is stored in a unified high performance storage layer, while compute resources scale elastically on demand. This reduces storage costs by more than thirty percent and allows AI training and inference workloads to allocate resources flexibly, which resolves resource contention between large and small tasks and establishes a solid foundation for implementing intelligent closed loop capabilities.

In a data infrastructure project for a municipal data bureau, the integrated Data and AI platform from KeenData enabled non algorithm teams to prepare data through the corpus processing layer and to complete model training, fine tuning and deployment with zero code on the intelligent support layer. They were then able to invoke APIs or build agents to transform large models into commercial products rapidly. The platform connected multimodal data with industry specific intelligent agents across the entire chain, covering the full lifecycle from data to models to applications, which supported rapid development of data products for targeted scenarios. Through standardized SDKs and plugin interfaces that allow third party corpus processing tools to integrate in a plug and play manner, the project accelerated the deployment of large models in urban service scenarios and promoted AI technologies that support city governance and industrial upgrading effectively.

In a Digital Government 2.0 project for another municipality, a trusted data space was built on KeenData’s integrated Data and AI platform, which provided a new generation of smart city big data infrastructure and trusted data environments. This enabled comprehensive digital, intelligent and refined management and services across government, public livelihood and industrial domains, while ensuring that data resources were deeply explored and quickly applied. The project established the first centralized data infrastructure support platform on the government side and explored effective mechanisms for providing public sector data to enterprises.

In a data intelligence infrastructure project for a major central state owned enterprise, the unified data center and governance framework were built on KeenData’s integrated Data and AI platform, which enabled efficient storage and computation for newly generated big data. By further aligning with business scenarios, the platform provided hundreds of service capabilities for planning, engineering decision making and integrated engineering platforms. With AI which drives the management and sharing of all business and research data, the project accelerated the transformation of data into digital resources and assets, which enhanced operational efficiency and achieved integrated operations across the business chain. This serves as an important milestone which marks the group’s transition into a new stage of highly coordinated intelligent operations.

Facing the opportunities and challenges of the data intelligence era, KeenData focuses on deep integration and innovation in Data and AI and is committed to building a solid foundation for digital and intelligent transformation for large organizations and enterprises. To date, KeenData has supported Data and AI infrastructure development for nearly two hundred large domestic and international organizations across more than twenty industries. These include Fortune 500 companies such as Sinopec, China Unicom, China Telecom, China CITIC Bank, AEON Group, China FAW and Sany Group, all of which build their Data and AI foundations on the KeenData Lakehouse platform. Looking ahead, KeenData will continue to drive industrial transformation through technological innovation based on KeenData Lakehouse, which will help large organizations and enterprises unlock new AI driven productive capabilities, seize competitive advantage in the digital era and jointly advance the industry toward a new stage of intelligent development.