Infini Founder Christian: The "personification expression" of large models has become the core barrier, and the moat of agents is shifting from models to application layers.

Crypto News, April 21st, during the roundtable discussion “Decoding Web 4.0: When AI Agents Take Over On-Chain Permissions,” Infini founder Christian stated that regarding the topic of “Agent Moats,” from a product and growth perspective, the core competitiveness of AI still primarily depends on the language processing capabilities of large models, especially the “human-like quality” of content output. He pointed out that compared to earlier models with mechanical and templated responses, current users are more concerned with whether AI can provide expressions and content quality close to real human conversation, which directly impacts its practical value in external communication and team collaboration. Christian mentioned that as model capabilities continue to evolve, the basic information processing ability is becoming more homogeneous. Once large models can handle about 90% of standardized tasks, relying solely on the models themselves will no longer create long-term barriers. The competition among AI Agents is gradually shifting from model capability to specific application scenarios and workflow design. In the future, truly differentiated capabilities will no longer be limited to simple information processing or desktop automation but will involve deep understanding of business processes in complex scenarios like finance and trading, enabling higher-level task execution and decision support.