New antibiotics are urgently needed to tackle the growing problem of drug resistance. To identify potential novel antimicrobials, Chu et al. have deployed bioinformatic modelling and chemical synthesis to generate natural product structures from gene clusters that are predicted to encode non-ribosomal peptides (complex secondary metabolites produced by bacteria) from the human microbiota. This led to the identification of humimycins, which were particularly active against Staphylococcus and Streptococcus bacteria in vitro. In a mouse model of methicillin-resistant S. aureus peritonitis, the humimycins potentiated β-lactam antibiotic activity and increased survival.