Scientists from Kyushu University have leveraged ProtGPT2, a protein language model, to generate novel small enzyme variants optimized for bioprocessing applications. This in silico approach identified functional motifs and produced amino acid sequences structurally and functionally akin to natural enzymes, demonstrating high novelty while preserving activity. The study signifies a leap forward in enzyme engineering, enabling rapid design and development of minimal enzymes critical for enhancing bioprocess efficiency and drug development.