<p>Supramolecular complexes of volatile components and cyclodextrin are widely used to improve the stability of volatile compounds. However, the quality assessment of these complexes remains challenging. In this study, paeonol/cyclodextrin complexes were prepared and characterized using FT-NIR spectroscopy. Using the NIR spectra of samples with different paeonol inclusion states (fully encapsulated, partially encapsulated, and unencapsulated), an SVM model was developed to classify the inclusion state, and a PLSR model was used to quantify the paeonol content. The SVM model achieved 100% accuracy, and the PLSR model had an R² &gt; 0.90 for both calibration and prediction sets. The Comprehensive Analysis of Single Spectral Acquisition (CASSA) method was first proposed in this work, which combines the two models to simultaneously realize qualitative and quantitative analysis using a single spectrum. This approach enables the rapid and reliable quality evaluation of supramolecular complexes and demonstrates the high potential of FT-NIR coupled with artificial intelligence.</p> Graphical abstract <p></p>