INTERVIEW: Shanghai Able Digital turns structured knowledge into AI-ready infrastructure

The knowledge technology provider is building traceable, callable knowledge assets through its platforms, as knowledge graphs become its largest revenue contributor

By Doug Young

As AI moves from general-purpose models into professional applications, competition is shifting toward domain knowledge, workflows and delivery. Shanghai Able Digital Science & Technology Co. Ltd. (2687.HK) is positioning itself at that application layer after nearly two decades of serving academic and research institutions.

The company converts academic materials, experimental processes and expert know-how into structured, traceable knowledge assets that AI can read and call on. Its products include Polymas, intelligent agents, discipline-specific models, and its Meta Graph and Harness engines.

Knowledge graphs have become the company’s fastest-growing business since 2023. In 2025, revenue from that business rose 68.5% to 573.5 million yuan ($84.6 million), or 59.2% of total revenue. Deliveries rose 122.4% to 10,386, while overall gross margin increased to 65.5% from 61.9%. Shanghai Able Digital has also announced framework partnerships with Alibaba Cloud and Volcano Engine, combining their models and cloud capabilities with its knowledge assets and academic scenarios.

CFO Crystal Cao spoke with Bamboo Works about the company’s AI strategy, knowledge assets and changing institutional needs.

Bamboo Works: Tell us more about Shanghai Able Digital’s Polymas platform. How does it differ from mainstream large language models, and how does it fit into your broader AI strategy?

Crystal Cao: General-purpose models are strong in language and broad reasoning, but professional fields also require authoritative, traceable and updated domain knowledge, plus an understanding of academic workflows. That’s the layer Polymas provides.

Polymas combines our domain AI with leading foundation models and structured specialist knowledge. Through the Meta Graph, answers can be linked to academic materials and knowledge nodes, making reasoning more transparent and reducing unsupported responses – what we call evidence-based AI.

Our strategy is model-agnostic: foundation model + knowledge assets + agents + application scenarios. This lets us adopt advances without relying on one provider.

Q: Your latest annual report describes knowledge graphs as the core growth engine of your digital knowledge content business. How does the product work, and how has it contributed to profitability?

A: Our knowledge graphs do more than visualize information. Academic knowledge is often dispersed across textbooks, papers, courseware, protocols and expert experience. We convert it into structured, traceable, updated and machine-callable assets.

Students can see how concepts and experiments connect. Teachers can generate materials, design assessments and update content from the same foundation. Institutions can map curricula and track scientific and industrial developments. For AI, the graph provides context, sources and guardrails.

This reflects a shift from knowledge storage to knowledge computation. Textbooks may take years to update, while knowledge graphs can be refreshed as research emerges.

At WAIC 2026, we introduced the Meta Graph and the Harness engines, which organize disciplinary knowledge into modular products for different scenarios. This is intended to support more reusable and standardized delivery alongside project-specific work.

Q: You recently announced partnerships with Alibaba Cloud and Volcano Engine. What does each side bring to these collaborations?

A: The technology companies bring foundation models, cloud computing and AI ecosystems. We bring nearly two decades of knowledge-infrastructure experience, multidisciplinary structures, real teaching and research scenarios, and the ability to turn them into deployable products.

A general-purpose model does not automatically understand disciplinary logic, experimental standards or institutional workflows. We act as a translator and knowledge anchor, converting general AI into professional teaching, research and training solutions. This institutional engagement is our real moat.

With Alibaba Cloud, the cooperation includes knowledge-token enablement, physical AI research and multi-agent platforms using the Qwen ecosystem and AgentScope. With Volcano Engine, it includes Doubao knowledge fine-tuning, integrated virtual-and-physical training and digital talent development.

We are also in exploratory discussions with other leading general-purpose AI providers about potential cooperation involving our knowledge assets, application scenarios and delivery capabilities.

Q: How do your products create value for administrators, teachers and students, especially for your key ‘lighthouse customers’? And what new products are you developing?

A: Our lighthouse customers are co-development partners. We begin with a high-value discipline or workflow, work with academic experts to build an authoritative knowledge foundation, connect it with teaching, research, assessment or experimentation systems, and turn the solution into reusable modules for other institutions.

For administrators, our products support evaluation and orchestration: structured evidence for learning outcomes, and tools for planning AI transformation and aligning curricula with technology and industry needs. For teachers, they turn materials and experience into reusable knowledge assets while reducing repetitive work. For students, the same foundation supports personalized learning, interactive simulations and source-traceable answers.

Our next generation of products centers on the Meta Graph and the Harness engines’ ‘1+N+1+N’ architecture: one knowledge foundation: multiple teaching, research, industry and experiment modules; one talent-development platform; and multiple discipline-specific configurations.

Q: What are academic and research institutions asking for most as AI becomes more widely adopted? And what opportunities do you see ahead?

A: The question has shifted from whether to adopt AI to how to integrate it responsibly into teaching and research. Institutions want trustworthy outputs, discipline-specific knowledge systems, integration with existing platforms, clear governance and security, and measurable improvements in outcomes.

Student expectations have also changed. Putting textbooks or lectures online is no longer enough. Students increasingly expect on-demand, interactive and personalized knowledge. They want AI to explain difficult concepts in different ways, connect prerequisite knowledge, simulate experiments, provide immediate feedback and show its sources.

At the same time, technology, industry practices and academic knowledge are evolving faster, requiring continuous curriculum renewal. This is driving a shift from resource digitization to knowledge operationalization: static content must become a dynamic system that can be updated, combined, used by people and called on by AI. We see this as a major structural opportunity.

Meanwhile, research may offer an even larger opportunity: AI can assist with literature review, hypothesis generation, experiment design, simulation and computation, data analysis and validation, and research output.

Q: Can you share a project that illustrates your capabilities in scientific research and knowledge services?

A: One example is our work in mechanics under China’s national ‘101 Plan,’ led in this field by Zhejiang University. The AIM mechanics model, a discipline-specific AI model for mechanics, is built on 2,026 knowledge points and 4,189 relationships, and connects them with a hypersonic wind-tunnel virtual simulation and a dedicated solver.

The project links concepts, equations, experiments, simulations and research tools into a discipline-level knowledge system. Students can move from a concept to an equation and then into a simulation, while teachers and researchers can update and reuse the same foundation as new results emerge. This helps research knowledge enter teaching faster.

Q: How are current policy priorities, including China’s 15th Five-Year Plan for 2026-2030, creating opportunities for your business?

A: We see three reinforcing forces. Higher education, scientific research and talent development are becoming more closely integrated with national innovation and industrial upgrading. New technologies require new disciplines, updated curricula and more complex practical training. At the same time, faster knowledge creation is increasing the operational burden on academic institutions.

The opportunity is therefore not simply a result of supportive policy. Institutions must absorb new research, update teaching content, redesign practical training and prepare teachers and students for emerging technologies much faster. Traditional, manually maintained content systems cannot keep pace.

Our products address that challenge by making knowledge structured, traceable, updatable and callable. The Meta Graph helps institutions understand and govern their knowledge, while the Harness engine applies it across teaching, research, experiments and talent development.

Demand currently exceeds our delivery capacity at a number of institutions, with some implementation schedules extending into 2027 and, in certain cases, 2028. This multi-year pipeline provides greater visibility into future delivery and supports continued growth, subject to project execution.

The Bamboo Works offers a wide-ranging mix of coverage on U.S.- and Hong Kong-listed Chinese companies, including some sponsored content. For additional queries, including questions on individual articles, please contact us by clicking here.

To subscribe to Bamboo Works‘ free weekly newsletter, click here.