DEEPEXI TECH Founder: General AI Models Struggle with Enterprise Pain Points; 'Ontology Large Model' Drives Subscription-based Transformation

DEEPEXI TECH (01384.HK) -4.200 (-6.061%) , the first enterprise-level large model AI application stock and an enterprise-level Token productivity platform company, saw its founder, Executive Director, Chairman and Chief Executive Officer Zhao Jiehui publish an article on artificial intelligence industry insights in mainland media today (14th). He pointed out that one of the core solutions to difficulties in enterprise AI implementation lies in the "ontology large model". This technology not only compensates for the lack of proprietary business logic in general models, but also transforms previously high-cost "manually customized" AI projects into standardized products, directly driving a shift in business models toward a "subscription-based" model. This provides a clear pathway for AI enterprises to move from "concept demonstration" to "creating tangible results."Zhao Jiehui noted that general models available in the market excel at semantic understanding but lack real internal business logic of enterprises. Taking the manufacturing industry as an example, two different automakers may encounter the same machine alarm, yet the underlying causes could be entirely different. Such proprietary troubleshooting experience and rules are not available in public online data or standard operation manuals, leaving general models often helpless when dealing with in-depth enterprise operations. What enterprises truly need are dedicated AI employees equipped with deep business planning capabilities.To break through this bottleneck, DEEPEXI TECH has launched the Deepexi Ontology Large Model tailored specifically for enterprises. The model is built on the vast dataset accumulated over the past eight years from serving nearly 400 large clients. This includes 108 "business ontologies" deeply covering key industries such as manufacturing, consumer retail, healthcare, finance and government affairs. These core operational data have never appeared in public corpora, successfully establishing a technological and data moat that pure model developers find difficult to replicate.In terms of commercialization, the ontology large model has enabled DEEPEXI TECH to achieve a leap in engineering economics. Zhao stated in the article that enterprise AI applications previously relied on frontline deployed engineers (FDEs) to conduct manual instantiation and customization on-site at client locations, similar to the deep engineering service model validated by US giant Palantir. Through the ontology large model, DEEPEXI TECH has successfully distilled and "modelized" these heavy tasks, significantly reducing customization costs per client and becoming a productization pioneer of this pathway in the China market.The Deepexi enterprise large model adopts a three-layer training architecture of "capability training industry ontology pre-training customer ontology injection." The first two layers, as fixed costs, are amortized across a large client base, enabling the customization cost per client to decline significantly with economies of scale. This has driven DEEPEXI TECHs transformation from a traditional project-based consulting model to a "Token consumption + AI employee man-hour subscription" model. The business model has shed its reliance on heavy customization and can expand linearly with scale, substantially shortening the profit cycle.Zhao concluded that while general models reduce Token unit costs on the cost side, the Deepexi Ontology Large Model enhances the business value density of Tokens, truly delivering economic benefits for enterprise AI applications. As a pure B2B enterprise-level large model AI application and enterprise-level Token productivity platform company, DEEPEXI TECH is leveraging this unique positioning and subscription model to achieve scalable monetization and continuously improve operational quality. (ad/da)(HK stocks quote is delayed for at least 15 mins.)