When "token" became the core theme of the 2026 China International Big Data Industry Expo, the answer KeenData has long been pursuing became increasingly clear: the foundation model is the engine, tokens are the fuel, and the quality and efficiency of that fuel are fundamentally shaped by AI data infrastructure.

Making its debut at the Expo, KeenData offered a comprehensive response to the "token era": the launch of Keen Agentic OS, the evolution of the KeenData Lakehouse Multi-modal Data Lakehouse Platform, and the transformation of data into understandable, callable and transferable AI production resources; its case was selected as a 2026 AI + Trusted Data Space construction practice achievement; from embodied intelligence to digital government, its ideas resonated across multiple forums; and the exhibition booth remained a hub of activity, where the trust accumulated through more than 300 organizations converged with real-world industry engagement. Product launches, recognition, forum discussions and roadshows formed a series of highlights. Here is a look back at KeenData's journey at the Expo.

Throughout the Expo, KeenData’s booth remained a focal point of activity. Representatives from the Guiyang municipal government, Shanghai Data Bureau, Tianjin Data Bureau, Zhengzhou High-tech Zone Administrative Committee and other government and related institutions visited the booth, gaining an in-depth understanding of KeenData’s technology roadmap and industry practices in AI data infrastructure, high-quality datasets, trusted data spaces and large-scale agent deployment.

Industry customers and partners from public resource trading, government affairs, energy, finance, manufacturing and other sectors also visited the booth, engaging in discussions on how AI can enter real business processes and how AI data can support models and agents. Authoritative media outlets including Guiyang Daily and Tianyan News also focused on KeenData, reporting on its product innovation and industry practices in AI data infrastructure.

Beyond the exhibition booth, KeenData participated in multiple roadshow activities, including Data Marketplace sessions, bringing its technology capabilities and industry practices into broader industry exchanges. The 2026 Expo was more than a technology showcase for KeenData; it was also a systematic expression of its vision for the era of AI industrialization.

On August 28, KeenData officially launched Keen Agentic OS, its intelligent agent development operating system, at the Expo booth, alongside the upgraded KeenData Lakehouse Multi-modal Data Lakehouse Platform.

As AI rapidly moves from foundation-model training toward large-scale agent deployment, challenges including data adaptation, task execution, system coordination, computing-resource scheduling and security governance remain critical for enterprise agents entering production environments. Built on extensive engineering practice, Keen Agentic OS integrates core capabilities including deep Data + Agent integration, ontology-based semantic knowledge bases, Agent Runtime, Multi-Agent collaboration, training and inference performance, and enterprise-grade governance and security. It enables agents to evolve from simply “being able to converse” to being able to understand, plan, execute and collaborate.

The upgraded KeenData Lakehouse further strengthens AI-oriented data supply capabilities. Through multi-modal data processing, cross-modal alignment, token extraction, data governance and training/inference acceleration, the two products together form the KeenData Agentic Lakehouse Platform, an intelligent foundation that connects the full chain from data to models and from models to agents.

KeenData also hosted a series of technical sessions, using product demonstrations, live showcases and case studies to further illustrate the technology behind scalable agent deployment. The key to AI industrialization is not simply having more powerful models, but building infrastructure that can continuously supply AI data, accumulate industry knowledge and support intelligent action.

During the Expo, KeenData participated in multiple core forums and industry exchange activities, engaging with industry stakeholders on embodied intelligence, AI industrialization, digital government, public resource trading and trusted data spaces.

At the Gui’an New Area Embodied Intelligence Data Element Industry Seminar and Investor Exchange, KeenData Chairman Yu Yang joined a closed-door roundtable discussion on data requirements across the full lifecycle of embodied intelligence. He emphasized that high-quality AI data supply is a critical foundation for embodied intelligence to move from laboratories toward large-scale applications.

At the Guizhou Digital Intelligence Industry Exchange, Cooperation and Supply-Demand Matching Conference, Vice President Zhu Jianyong delivered a keynote speech titled “New Infrastructure for the Era of AI Industrialization.” Centered on the “data factory” and “model factory” concepts, he proposed continuously producing data and continuously training and validating AI capabilities to build a new production system for industrial intelligence.

At the Digital Government themed forum roundtable, Vice President Zhu Jianyong participated in discussions under the theme “Empowering High-Quality Digital Government Development through Digital-Intelligence Collaboration.” He noted that digital government is moving from data aggregation toward data intelligence, with the key challenge being the development of AI-era data infrastructure that enables data to become productive resources that models and agents can understand and call upon.

At the Digital-Intelligence Public Governance and Public Resource Trading Big Data Exchange, Vice President Guo Zhenqiang delivered a keynote titled “Data Restructuring: New Infrastructure for the Era of AI Industrialization.” He proposed that “the first half of AI is model breakthroughs, while the second half is the restructuring of industrial systems,” and outlined a three-layer transformation from data that can supply AI, to data that accumulates industry knowledge, and ultimately to agents that drive industrial action.

Vice President Gao Hailing participated in a closed-door discussion at the Public Resource Trading Data Quality Improvement and Analytical Application Seminar. Vice President Zhu Jianyong also joined a roundtable at the AI + Trusted Data Space Development Exchange, discussing data standardization, data quality improvement, trusted data circulation and AI-driven trusted data space development with industry representatives.

Different forums addressed different industry scenarios, but the discussions ultimately converged on the same direction: as AI moves deeper into industry, AI data infrastructure is becoming a critical foundation connecting data, models, agents and real-world business.

During the Expo, KeenData’s case “Urban Trusted Data Space: Building a Foundation for Urban Data Element Circulation” was selected as a 2026 AI + Trusted Data Space Construction Practice Achievement. Powered by the self-developed KeenData Agentic Lakehouse Platform, KeenData participated in building a trusted and controllable infrastructure covering the full lifecycle of data circulation and utilization, exploring new approaches to secure and compliant public-data supply, high-quality dataset development and market-oriented allocation of data elements.

This recognition is not only an endorsement of an individual practice case, but also a response to KeenData’s long-term commitment to AI data infrastructure and its continued efforts to integrate data elements with artificial intelligence. From data governance and trusted circulation to high-quality dataset development and AI applications, the value of technology ultimately has to return to real-world scenarios and be validated through industry practice.

The Expo has come to an end, but the exploration continues.

As model capabilities increasingly become foundational, the real competition is shifting toward whether industrial systems can be fundamentally restructured. Whether data can be continuously supplied, knowledge can be effectively accumulated, and agents can enter business processes will determine whether AI can truly become a productive force for industry. In the next stage of AI industrialization, KeenData will continue to deepen its work in AI data infrastructure, transforming data from resources into intelligence, and intelligence from technical capability into real industrial value.