Two complementary reports in Nature Biotechnology describe how deep‑learning models trained on high‑throughput screening and phenotypic data accelerated discovery of novel antibacterial compounds. Teams combined phenotypic screening with neural networks to prioritize chemically distinct scaffolds that show activity against resistant bacteria in follow‑up assays. Authors and journals note the approach reduces the search space for medicinal chemistry and could speed preclinical candidate selection against multidrug‑resistant pathogens.