An AI system was designed to measure endometrial thickness and its performance was compared with expert clinicians across multicenter IVF ultrasound datasets. The study demonstrated that AI achieved accuracy and consistency on par with experienced sonographers, with variability nearly identical to clinician-to-clinician differences. In clinically relevant decision-making, such as determining whether thickness exceeds the 7 mm threshold, the AI matched or slightly outperformed human agreement levels. These findings highlight the system’s potential to standardize ultrasound interpretation, reduce observer-dependent variability, and support more reliable decision-making in IVF. By improving reproducibility across centers, this technology can also strengthen large-scale research and streamline clinical workflows.