With CO2 emissions destroying the environment, cheap ways to separate the compound from related gas mixtures are regarded as one of the biggest environmental challenges of our century. One alternative to the current methods is membrane-based separations. The EU-funded SmartDeZIgn project provides a novel method for the design of zeolitic imidazolate frameworks (ZIFs) for CO2 selective membranes. Experts aim to develop a computational tool based on machine learning methods that will screen all the suitable metals and linkers in combination with the hundreds of available ZIF topologies.