The generative rearchitecture of antibody engineering shifts empirical discovery into intentional design

Abstract Generative AI is reshaping antibody engineering by transitioning from empirical screening toward more specification-driven design, potentially expanding the accessible target landscape. While front-loading developability criteria helps streamline discovery timelines, data constraints still introduce risks of in silico misprioritization. This Perspective discusses how integrating computational inference within disciplined engineering frameworks can complement traditional sampling methods, highlighting that coupling AI-driven predictions with rigorous empirical validation remains essential for achieving therapeutic intent.