Researchers published an atomically accurate, de novo antibody design method using RFdiffusion that produced antibodies with confirmed atomic binding poses by cryo‑electron microscopy. The Nature paper reports AI-generated single-domain antibodies (VHHs) able to target specified epitopes and the work highlights how computational design can bypass lengthy experimental screening. The paper drew commentary from the field, including remarks by Nobel laureate David Baker, underlining both the promise and early-stage nature of computational antibody design. Cryo‑EM validation across multiple therapeutic targets strengthened claims of precise epitope engagement but the authors noted downstream optimization and developability remain necessary steps for clinical translation. The work accelerates the industry’s push to embed generative protein design into biologics pipelines; biotechs and large pharma will test whether atomically accurate starting points reduce cycle time and cost for therapeutic antibody programs.