Machine Learning‐Driven Automated Synthesis of Polysubstituted Gentisaldehydes

Gentisaldehyde is a fundamental motif with broad applications in pharmaceuticals. However, the construction of such molecules is time- and cost-intensive with low step efficiency. The development of a disruptive retrosynthetic method is therefore highly necessary to enhance ring-formation capability and reduce redundancy in synthetic strategies. Herein, guided by cutting-edge computer-aided synthesis planning (CASP) algorithms, synthetic routes were systematically deduced toward polysubstituted gentisaldehydes (PGAs). An automated flow system was subsequently developed to implement the streamlined synthesis via selective 6-endo cyclization of cyclobutenedione derivatives and propargyl diacetal moieties. DFT computational studies further revealed the involvement of diradical intermediates, rather than the conventional zwitterionic mechanism, and identified the 1,5-hydrogen atom transfer process as the key driving force. The rapid and collective construction of a gram-scale library of PGAs highlights the industrial potential for scalability and application. This work demonstrates the synergistic interplay among computational retrosynthetic analysis, mechanistic elucidation, and flow processing, establishing an innovative model for universally integrated molecule library construction.