Abstract: The present invention provides a computer system for generating the molecular structure of a catalytic activator for a reaction in which input reactants, a.k.a. substrates, are converted into an output product, the computer system comprising: a trained machine learning model, preferably a variational autoencoder, configured to receive an operating feature set defining chemical features of the input reactants and chemical features of the output product of a reaction and to generate therefrom a set of catalyst features defining one or more catalytic activator, which is preferably an enzyme, for catalysing a reaction to convert the input reactants to the output product.