Non-natural cyclic peptides constitute an underutilized yet attractive segment of the chemical diversity space. Such peptides are conformationally, chemically and proteolytically stable, the incorporation of non-proteinogenic amino acids generally lowers their immunogenicity, their binding interfaces can be of the size and complexity not achievable in small molecules, and all these desirable properties materialize in relatively short sequences that can be readily synthesized using the routine solid phase methodology. Such artificial peptides cannot be, by definition, derived from natural sequences. Combinatorial chemistry faces substantial difficulties in this case due to the combinatorial complexity involved. De novo design is thus the preferred enabling technology for this class of molecular entities. An original variant of the Monte Carlo chain growth algorithm was applied to the design of short cyclic peptides - CD4-mimicking ligands of HIV gp120. In the role of entry inhibitors, the engineered cyclic peptides performed by 1 to 2 orders of magnitude better than the mechanistically similar small molecule positive standard NBD-556. Conclusion: Structure-based de novo design is capable of delivering cyclic peptides that fill the gap between small molecules and biologics. The case study indicates a satisfactory capability of these synthetic compounds to mimic biologics, while offering much better drug-like properties.