Artificial Intelligence-Driven\nDevelopment of Nickel-Catalyzed\nEnantioselective Cross-Coupling Reactions

The conventional approach to developing\nasymmetric synthetic\nmethods\nrelies heavily on empirical optimization. However, the integration\nof artificial intelligence (AI) and high-throughput experimentation\n(HTE) technology presents a paradigm shift with immense potential\nto revolutionize the discovery and optimization of asymmetric reactions.\nIn this study, we present an efficient workflow for the development\nof a series of nickel-catalyzed asymmetric cross-coupling reactions,\nleveraging AI and HTE technology. Many nickel-catalyzed enantioselective\ncross-coupling reactions share a common Ni(III) intermediate, which\ndictates the enantioselectivity. To harness this mechanistic insight,\nwe embarked on developing a predictive model for nickel-catalyzed\nenantioselective coupling reactions, elucidating the general rules\ngoverning enantioselectivity. Through the application of data science\ntools and HTE technology, we curated a data set to construct an AI-based\nmodel. This model was subsequently utilized to facilitate the discovery\nof efficient nickel hydride-catalyzed enantioselective and regioselective\ncross-coupling reactions. Employing AI-assisted virtual ligand screening\nand HTE-enabled condition optimization, we successfully identified\noptimal ligands for eight coupling reactions. Consequently, a series\nof chiral sp<sup>3</sup> C–C bonds were synthesized with high\nyield and enantioselectivity.