Glycosylation is an effective means to alter the structure and properties of plant compounds, influencing the pharmacological activity of natural products (NPs) to obtain highly active NPs. In nature, glucosides are the most widely distributed, while other glycosides such as xylosides are less common and present in lower quantities. This is due to the scarcity of xylosyltransferases with substrate promiscuity in nature, and the modification of their catalytic function is also quite challenging. In this study, we first performed a phylogenetic analysis of reported UDP-glycosyltransferases (UGTs) of plant and microbiological origin and identified a unique motif region from the UGTs of the Bacillus genus, which may be responsible for the broad sugar donor catalytic activity of the UGTs in the Bacillus genus. Then, utilizing protein engineering techniques, we have evolved a xylosyltransferase M3-2, which exhibited high substrate promiscuity, sugar donor promiscuity, and site selectivity, enabling the synthesis of a variety of O-glycosides. In addition, another mutant M3-1 has been engineered to alter the sugar donor specificity of the UGT, enabling the switch from UDP-Glc donor to UDP-Xyl. The improved enzymatic activity is likely attributed to stable hydrophobic interactions and hydrogen bonding interactions between the enzyme and the substrate. In order to synthesize xylosylated products more economically and efficiently, an in vitro synthetic pathway that utilizes NPs and inexpensive glucuronic acid as starting materials was designed. Through this pathway, we successfully synthesized a variety of unnatural xylosylated products belonging to O-glycosides, one of which 10a possesses excellent anti-inflammatory activity. We anticipate that this work will contribute to the future discovery and industrial production of unnatural glycosides.