•Selection and modeling of AMPs derived from Lippia transcriptome.•Peptides were active against both gram-positive and gram-negative bacterial strains.•Hemolysis was mitigated through rational design of the peptides.•Successful selection of AMPs was modeled from Lippia transcriptome.•Former peptides were active against Gram-positive and Gram-negative bacteria.•Rational design reduced hemolytic activity in blood cells. Antimicrobial resistance is considered a health issue worldwide. This public health problem underscores the importance of searching for new antimicrobial molecules with different mechanisms of action. Leaf transcriptomes were used to search and develop synthetic antimicrobial peptides derived from mRNA sequences. The in silico search for new AMPs from the L. rotundifolia and L. alba transcriptomes allowed the identification of 120 putative peptide mRNA sequences. Eight of them fitted into optimal parameters and were translated and chemically synthesized antimicrobial peptides. Their biological activity was tested in both Gram-positive and Gram-negative bacteria against which they exhibited antibacterial activity. However, they showed an important hemolytic effect. Afterwards, two active peptides showing bactericidal activity isolated from each plant transcriptome tested were modified and modeled in 11 new variants to increase their antimicrobial activity and stability and to reduce or eliminate their hemolytic effect from their original peptides. The La-AMP1 (MSLLERKLLMHFLRV) the original peptide from L. alba showed a 52% hemolytic effect while the derived peptide La-AMP1a (GLMKLLRELLHMFSRVG) had its hemolytic effect reduced to 0.5% at 128 μg.mL-1. Similarly, we observed that the original peptide from L. rotundifolia, Lr-AMP1 (MRIGLRFVLM), displayed a 71.5% hemolytic effect, while its derived peptide Lr-AMP1f (GSVLRAIMRMFAKLMG) showed 0% hemolysis at 128 μg.mL-1, tested with fresh human erythrocytes. Our results indicate a promising method for the search for novel antimicrobial agents with reduced or zero hemolytic effect, as well as prediction and optimization of their activity from plant mRNA libraries.