Bacterial tyrosine kinases (BY kinases) control the synthesis and export of capsular polysaccharide in several major human pathogens, and are required for immune evasion and biofilm formation. They have no counterpart in the human proteome, although their P loop nucleotide binding fold is widespread among human NTPases, so selectivity must be demonstrated rather than assumed. Work in our group has shown that peptide deformylase binds and directly inhibits the Bacillus subtilis kinase PtkA, establishing that this nucleotide binding site can be blocked by a peptide. This project designs peptide inhibitors of four BY kinases: PtkA (Bacillus subtilis), CapB2 (Staphylococcus aureus), CpsD (Streptococcus pneumoniae) and Wzc (Klebsiella pneumoniae). Three of the four have no experimental structure and are first modelled with AlphaFold2. Candidate binders are generated with RFdiffusion3, sequences are assigned with ProteinMPNN, and complexes are filtered with ESMFold and by AlphaFold2 initial guess prediction. Designs passing the interface confidence threshold are screened for cross reactivity against human NTPases, then passed to collaborators for peptide synthesis, inhibition assays and molecular dynamics. Experimental results feed back into successive design cycles.