Basic leucine-zipper (bZIP) proteins represent difficult yet compelling oncogenic targets, since numerous cell-signalling cascades converge upon them where they function to modulate transcription of specific gene targets. bZIPs are widely recognised as important regulators of cellular processes that include cell proliferation, apoptosis and differentiation. Once such validated transcriptional regulator, Activator Protein-1 is typically comprised of heterodimers of Fos and Jun family members, with cFos-cJun being the best described, and demonstrated to be key in the progression and development of a number of different diseases. As a proof-of-principle for our approach, we describe the first use of a novel combined in silico/in cellulo peptide-library screening platform that facilitates the derivation of a sequence which displays high selectivity for cJun relative to cFos, while also avoiding homodimerisation. In particular, >60 million peptide peptides were computationally screened and all potential on/off targets ranked according to predicted stability, leading to a reduced size library that was further refined by intracellular selection. The derived sequence is predicted to have limited cross-talk with a second previously-derived peptide antagonist that is selective for cFos in the presence of cJun. The study provides new insight into the use of multi-state screening with the ability to combine computational and intracellular approaches in evolving multiple co-compatible peptides that are capable of satisfying conflicting design requirements.