A natural deep eutectic solvent-based sonication assisted liquid-phase microextraction (NDES-SA-LPME) combined with UV-Vis spectrophotometry was optimized for the analysis of thiamine (Vit-B1) in dairy products, fruits, nuts and vitamin tablets. Five NADES was prepared by blending different molar ratios of hydrogen bond acceptors (citric acid, malic acid, tartaric acid, choline chloride and betaine) and hydrogen bond donors (glucose, proline, and glycerol). Based on preliminary experiments, the central composite design was created for optimization of selected key extraction factors. The effects of factors on the extraction of Vit-B1 were investigated using artificial intelligence approaches. Validation of the NDES-SA-LPME method was performed by analysis of reference materials. Working ranges for model solutions and matrix matched solution were 8–500 µg L−1 and 15–400 µg L−1, respectively. Further, limits of detection for model solutions and matrix matched solution were 2.4 µg L−1 and 4.6 µg L−1, respectively. Finally, a matrix matching calibration approach was used to increase recovery results, and acceptable recovery data were obtained in range of 95.9%-118.4%. This study is the first report to apply artificial intelligence, experimental modelling and NDES together to the analysis of Vit-B1 in real samples.