Sufficient confinement of electromagnetic modes leads to a frequent exchange of energy between a material and light, resulting in the creation of hybrid light-matter states, called polaritons. Here, I illustrate how spectral design with polaritons is utilized to reach mode- and enantiomer-selective catalysis. We discuss that strong interaction between optical and vibrational modes leads to changes in chemical rate constants [1-2]. Machine learning is critical to tame the computational complexity of experimentally relevant systems. Furthermore, a proof-of-concept work demonstrates that meta-material design can be used to boost charge transfer at plasmonic surfaces [3] -- promising a bright future for polaritonically driven catalysis guided by machine learning. [1] C. Schäfer, J. Flick, E. Ronca, P. Narang, and A. Rubio, Nature Communications, (2022) 13:7817. [2] C. Schäfer, J. Fojt, E. Lindgren, and P. Erhart, J. Am. Chem. Soc. 2024, 146, 8, 5402-5413. [3] J. Fojt, P. Erhart, C. Schäfer, Nano Lett. 2024, 24, 38, 11913–11920.