Intelligent Software Application Ecosystem: Paradigm Shift and Engineering PathwaysReport of the 34th CCF Beautiful Lake Seminar

Artificial intelligence is moving from a phase of “tool empowerment” toward deeper integration with software systems. AI technologies such as large language models and agents are fundamentally reshaping the form and construction of software. Based on the discussions from the 34th CCF Beautiful Lake Seminar on “Intelligent Software Application Ecosystem”, and incorporating technological advancements in the nearly nine months following the seminar, this article systematically reviews the core issues and development trends of the intelligent software application ecosystem. Starting from the capability boundaries of large language models and the paradigm shift in intelligent software, we explore engineering approaches for intelligent software, trace the evolutionary trends of AI engineering practices from harnessable agent systems to dynamic optimization and then to organizational collaboration, and finally propose a development roadmap for enterprise-grade software centered on neuro-symbolic integration as the architectural core and deconstruction-reconstruction as the engineering path, aiming to provide a reference for both technological innovation and industrial deployment of intelligent software.