Nonribosomal peptide synthetases (NRPSs) assemble bioactive peptides from an enormous repertoire of building blocks. How binding pocket residues of the nonribosomal adenylation domain, the so-called specificity code, determine which building block becomes incorporated has been a landmark discovery in NRPS enzymology. While specificity codes enable the prediction of substrate specificity from protein sequence, design strategies based on rewriting the specificity code have been limited in scope. An important reason for failed NRPS design has been that multispecificity has not been considered, for a lack of suitable assay formats. Here, we employ a multiplexed hydroxamate specificity assay (HAMA) to determine substrate profiles for mutant libraries of A-domain in the termination module the SrfAC of surfactin synthetase. A generalist version of SrfAC is developed and the functional flexibility of the adenylation reaction is probed by fully randomizing 15 residues in and around the active site. We identify mutations with profound impact on substrate selectivity and thus reveal a remarkable evolvability of A-domains. Statistical analysis of the specificity divergence caused by point mutations has determined the impact of each code position on specificity, which will serve as a roadmap for NRPS engineering. The shortness of evolutionary pathways between NRPS specificities explains the rich natural substrate scope and suggests directed evolution guided by A-domain promiscuity as a promising strategy.