Two complementary papers published in Nature Biotechnology demonstrate that deep‑learning approaches can materially accelerate antibacterial discovery by prioritizing novel scaffolds from high‑throughput and phenotypic screening datasets. One study combined phenotypic screening with neural network‑driven virtual screens to nominate compounds with antibacterial activity; a second paper trained models on experimental HTS readouts to surface structurally novel antimicrobial scaffolds. Both groups validated hits experimentally and highlighted AI as a route to overcome limitations of traditional screening libraries. Phenotypic screening remains the gold standard for whole‑cell activity; these reports show modern ML models can triage large chemical space more efficiently, reducing hit‑finding timelines.