It is a pressing need to develop new energy materials to address the existing energy crisis. However, screening optimal targets out of thousands of material candidates remains a great challenge. Herein, an alternative concept for highly effective materials screening based on dual-atom salphen catalysis units, is proposed and validated. Such an approach simplifies the design of catalytic materials and reforms the trial-and-error experimental model into a building-blocks-assembly like process. First, density functional theory (DFT) calculations are performed on a series of potential catalysis units that are possible to synthesize. Then, machine learning (ML) is employed to define the structure-performance relationship and acquire chemical insights. Afterward, the projected catalysis units are integrated into covalent organic frameworks (COFs) to validate the concept Electrochemical tests confirming that Ni-SalphenCOF and Co-SalphenCOF are promising conductive agent-free oxygen evolution reaction (OER) catalysts. This work provides a fast-tracked strategy for the design and development of functional materials, which serves as a potentially workable framework for seamlessly integrating DFT calculations, ML, and experimental approaches.