Generality‐Driven Optimization of Enantio‐ and Regioselective Mono‐Reduction of 1,2‐Dicarbonyls by High‐Throughput Experimentation and Machine Learning

Abstract The longstanding quest for substrate generality stems from the unpredictability of single‐model optimization. Leveraging high‐throughput experimentation (HTE), we present a practical multi‐substrate screening strategy for the general asymmetric mono‐reduction of 1,2‐dicarbonyls. Quantitative 1 H NMR spectroscopy combined with simultaneous chiral analysis by 19 F NMR for pooled crude mixtures accelerated the workflow eightfold. Robust screening of 31 chiral oxazaborolidinium ion (COBI) variants across eight substrates tackled even ethyl/methyl differentiation. HTE data were utilized in a machine learning (ML) model with CGR (Condensed Graphs of Reaction)‐based descriptors, identifying catalysts for target substrates without quantum chemical calculations. The ARMS (Automated Reaction Mapping for various Substituents) system was introduced to streamline SMILES (Simplified Molecular Input Line Entry System) preprocessing for multi‐substrate datasets. The resulting chiral α ‐silyloxy ketones, obtained in excellent yields (up to >99%) and selectivities (up to >99% ee, >20:1 r.r.), could be readily transformed into high‐value compounds, such as ( S )‐bupropion.