Researchers introduced SMRTnet, a deep-learning platform that predicts small-molecule–RNA interactions without requiring RNA tertiary structures. The model fuses large language models with graph and convolutional networks to infer binding using RNA secondary structure and chemical features. SMRTnet outperformed existing tools across benchmarks and identified candidate compounds against ten disease-associated RNAs, with several hits validated in vitro. One validated small molecule downregulated MYC IRES activity and suppressed proliferation in cancer cell lines, illustrating the platform’s potential to expand druggable RNA targets and accelerate discovery where high-resolution RNA structures are unavailable.