WAIC 2026 | ABLE DIGITAL (02687) debuts source map + Harness engine, the real bottleneck for high-value AI scenarios is "knowledge" not "models"

On July 18, 2026, at the World Artificial Intelligence Conference Industry Ecology Forum, ABLE DIGITAL (02687) was invited to speak as a benchmark enterprise in the field of knowledge technology and officially released the company's core achievements in the field of knowledge technology - the "Source Knowledge Graph" knowledge base and the new Harness engine. This was not just a regular technical release, but rather a systematic response by the company to the core pain points in the AI + knowledge industry: the bottleneck is not the model, but knowledge. It is worth noting that in the "Principals Dialogue" session of the same conference, seven academic presidents from Tsinghua University, Peking University, Shanghai Jiao Tong University, Huazhong University of Science and Technology, Harbin Institute of Technology, and The Hong Kong University of Science and Technology (Guangzhou) discussed "What should universities do when machines can 'teach'?" The key word that repeatedly appeared was "knowledge." ABLE DIGITALs release, precisely addressed this issue. I. Seven academic presidents jointly questioned, and ABLE DIGITAL's response was very direct During the forum, Peking University President Ding Kuiling said that the students' "head-up rate" was not high, not because they were not diligent, but because when AI can access any information at any time, the traditional "lecture-style" knowledge transmission has become ineffective. Huazhong University of Science and Technology President You Zheng was even more direct: AI can be a "scaffold," but not a "crutch." These discussions all point in the same direction: in the AI era, the core issue for universities is shifting from "how to teach knowledge" to "what knowledge to teach," as well as, who will provide this knowledge. Following this question down a level, a more practical issue emerges: if machines are going to "teach," how can they teach accurately? ABLE DIGITAL's judgment is very direct - the real pain point in the industry is not the model, but knowledge. The training data for general large models mainly comes from publicly available data on the internet, but the real knowledge that constitutes barriers to a discipline - textbooks, experimental data, decades of accumulated teaching experience, professors' professional judgments - are not easily accessible on the internet. For scenarios with high precision requirements such as physics experiment procedures, engineering design standards, medical diagnostic logic, factual illusions are not just "flaws", but "fatal flaws." II. Knowledge Base: ABLE DIGITAL allows every reasoning of AI to be verifiable ABLE DIGITAL first systematically unveiled its AI's underlying architecture - the "Source Knowledge Graph" knowledge base at WAIC. The "source" in the name points to a core demand: tracing back. On this knowledge base, the reasoning process of AI is no longer a "black box," and each step of deduction can be traced back to specific academic literature and knowledge nodes. The company serves nearly two thousand universities and academic research institutions, covering all disciplines and all knowledge has been verified and supports full chain traceability. These data come from teaching environments, experimental settings, and frontline research, with a high update frequency, deep structuring, and dynamic iteration. Essentially, the company is systematically transforming the implicit disciplinary knowledge that was initially scattered across textbooks, experimental specifications, teaching materials, and teacher experience, into a and structured knowledge infrastructure that is machine-readable and callable. III. Harness Engine "1+N+1+N": ABLE DIGITAL turns knowledge infrastructure into products Besides judgment, ABLE DIGITAL also released a new Harness engine - "1+N+1+N". The first "1" is the Source Knowledge Graph knowledge base, which breaks down disciplines into four layers of knowledge graph, capability graph, experimental graph, and evaluation graph, forming a discipline-level "knowledge operating system." "N" is the module of multiple scenario paradigms, covering different business scenarios such as research, teaching, industry, and experiments. The second "1" is the training platform, which undertakes the entire talent cultivation process. The last "N" is disciplinary customization, aimed at specific disciplines for deep adaptation. More crucially, the delivery method is: can be dismantled, can be combined, can be accessed on demand. Users do not need to overthrow their existing systems and rebuild, but can gradually connect through a modular access mode, integrating step by step by discipline and by scenario. This means that "1+N+1+N" is essentially a commercially viable knowledge infrastructure delivery solution - with discipline-level professional knowledge base and vertical models as the foundation, with features of scalable replication and decreasing marginal costs. At WAIC 2026, ABLE DIGITAL provided not only a product release, but also a systematic solution to the pain points faced by AI4S. Discipline-level knowledge data is the foundation, self-developed AI capabilities are the engine, and a nationwide service network is the access channel - the combination of the three allows "knowledge infrastructure" to move from concept to practical application. With the Source Knowledge Graph and Harness engine in more universities, the commercialization curve of the company's knowledge technology business is expected to accelerate, creating a thicker and more solid industrial moat for long-term value accumulation.