Researchers published a Nature paper demonstrating RFdiffusion, an AI‑driven protein design method, can generate de novo single‑domain antibodies (VHHs) that bind user‑specified epitopes with atomic precision; cryo‑EM confirmed binding poses for multiple therapeutic targets. The work shows AI can now design full‑length antibody scaffolds and hypervariable loops that historically required extensive experimental screening. Nobel laureate David Baker, co‑author and leader in protein design, commented on progress and caveats: while the model delivers proof of concept and structural validation, widespread adoption will require further validation across targets and integration into downstream development workflows. The research accelerates the prospect of rapid in silico antibody discovery that could shorten early discovery timelines. Why it matters: atomically precise AI design reduces reliance on labor‑intensive screens and could reshape target‑to‑lead timelines for antibody therapeutics; companies and investors are already reacting to commercial opportunity.