To tackle the urgency of smarter crop management, the complex nature of agricultural ecosystems needs to be better understood, employing and combining different techniques and technologies. In this study, untargeted metabolomics and agro-meteorological survey were coupled to study the variation of Sorghum bicolor (L.) Moench metabolome during crop development, in response to environmental and anthropic factors. Twelve crop fields in the Emilia-Romagna region, Italy, were monitored and sampled at different stages, seedling (Ss), advanced vegetative (Sv), and ripening (Sr), and subjected to 1H NMR-based metabolomics. The analytical method developed resulted to be successful to quickly analyze different sorghum organs. Dhurrin, a cyanogenic glucoside, resulted to be a biomarker of crop quality and development, and several insights into its turnover and functions were obtained. In particular, p-glucosyloxy-2-hydroxyphenylacetic acid was identified, for the first time, as the main metabolite accumulated in sorghum at Sr, after gradual dhurrin neutralization. During plant life, fertilization and biotic and abiotic stress reflected peculiar metabolomic profiles. Water supply and soil features (i.e., clay content) were correlated to metabolomic variations, affecting dhurrin (and related metabolites), amino acids, organic acids, and carbohydrate content. Increase in chlorogenic acid was registered in consequence of predator attacks. Moreover, grain from three fields presented traces of dhurrin and the lowest antioxidant potential, which resulted in poor grain quality. Metabolomics turned out to be a promising tool in view of smart agriculture for monitoring plant growth status and applying appropriate agricultural practices since the early stage of crop development.