Quantifying small molecules inside a cell is a major challenge in biology. Riboswitch-based biosensors promise to solve this by converting small molecule concentrations into directly detectable outputs, however, their potential has not been fully realised due to a lack of operational range. This thesis developed, tested, and used a systematic framework to characterise and evolve riboswitches from very large collections of randomised RNA sequences. A novel experimental approach was developed based on single cell fluorescence measurements and amplicon sequencing to both identify and validate improved riboswitch candidates.