Generative AI tools have provided much amusement for consumers, yet the same enthusiasm has not translated to the enterprise market. This stems from significant limitations in the large language models (LLM) behind generative AI applications, leading to highly fragmented AI agent applications, according to Taiwan industry observers, who suggest using a quadrant-like framework to define the necessity of AI tasks.