Modern antenna design is a daunting task aimed at fulfilling diverse performance requirements and constraints imposed by specific application areas. Traditional techniques are heavily based on engineering experience. This limits the options considered for conventional architectures and leads to sub-optimal results. This study introduces an innovative approach to automated development of antennas. The introduced methodology incorporates computational intelligence techniques and numerical optimization procedures to carry out unsupervised antenna topology generation and fine-tuning its geometry parameters. The crucial component of the suggested method is a versatile parameterization involving elliptical-shaped patches and gaps, which can realize a massive variety of different shapes. Computational intelligence is used to execute a purely specification-driven antenna evolution process. The decision variables include a mixture of discrete and continuous parameters handled by a customized evolutionary algorithm (Stage I) and local optimizer (Stage II – fine tuning). The procedure is fully specification-based and requires no human-expert interaction whatsoever. The proposed framework has been comprehensively demonstrated by designing several devices of distinct characteristics (broadband, multi-band, compact). The findings underscore the versatility of the technique and its suitability to produce nonconventional structures with acceptable computational costs.