Kingsoft Office Zhu Yi'e: From "Seeing" to "Mastering": AI Applications Enter the Era of "Knowledge-Enhanced Generation" | Alpha Summit

Zhu Yi'e pointed out that the current core bottleneck of AI applications is no longer the model itself, but how to transform the vast, unstructured private domain data within enterprises into high-quality knowledge that can be understood by AI. Traditional RAG is limited by "documents ≠ knowledge" and "semantic similarity ≠ logical relevance." Therefore, it is necessary to shift AI applications from "model-centric" to "data/knowledge-centric," developing "Knowledge-Augmented Generation" (KAG), driven by both data lakes and knowledge lakes, to govern, model, and utilize knowledge assets, allowing AI to truly "master" rather than just "see" enterprise knowledge, thereby achieving reliable efficiency improvements in scenarios such as professional Q&A and intelligent writing