Why Enterprise AI Needs Ontology Modeling

创业邦 · wechat · 2026-07-16

This article discusses the role of "ontologies" in enterprise AI, explaining why many large models that perform brilliantly in consumer apps often fail to understand business contexts on the ground. The author traces this back to the fact that enterprise knowledge, processes, roles, rules, and data have long been siloed across different systems, lacking a unified, low-ambiguity format that AI can reliably comprehend.

Citing multiple surveys, the article notes that enterprise AI currently delivers limited ROI and scalability. IBM's 2025 CEO survey shows only 25% of AI projects achieve expected returns, with just 16% reaching enterprise-wide scale. Meanwhile, McKinsey reports that while 78% of enterprises use generative AI in at least one business function, over 80% say it hasn't materially impacted profitability.

The author suggests that enterprise AI is evolving from search and Copilots to Agents, and ultimately toward an "ontology-based enterprise AI operating system." The article highlights case studies from Aihua Tech and Qiaoyin Shares: by leveraging ontology modeling and FDE teams, ticket response times dropped from hours to minutes, and idle driving rates for operational vehicles fell from 35% to 8%. Additionally, the upcoming implementation of the GB/T48000.3—2026 standard for ontology modeling on August 1 is seen as a major step toward industry standardization.

Original post →

More from Companies & People

Companies & People channel →