This paper introduces a novel machine learning-based methodology to determine the operational range of planar microstrip antennas of randomly generated designs, removing the need for electromagnetic (EM) simulations or expert knowledge. Framed as a multi-label classification task, the proposed approach addresses the inefficiencies of traditional methods, which are prone to high computational cost and engineer’s bias. The method quickly identifies promising designs, paving the way for subsequent optimization. This advancement represents a significant step toward automating antenna design processes.