The EMBRACE (Empirical data-Based crew Reliability Assessment and Cognitive Error analysis) method was developed to capture potential cognitive errors in procedure-following activities and failure possibilities due to time constraints. Grounded in an empirical foundation derived from full-scope simulator observations, this method aims to provide highly realistic human reliability assessment (HRA) results. However, the inherent complexity of HRA workflows, which require decomposing procedures into primitive tasks, often poses usability challenges that may induce human errors during the analysis process itself. To address this issue, the EMBRACE Support Software (ESS) was developed to assist analysts by automating the linkage between procedure sentences and primitive tasks and by visualizing the timeline and procedural flows of given human failure events (HFEs). This study presents a usability evaluation of the ESS, employing a hybrid approach that integrates AI-driven analysis via an AI-powered design intelligence platform with manual expert evaluations, specifically heuristic evaluation and cognitive walkthrough. By leveraging specialized AI agents for initial screening and human experts for contextual cognitive analysis, we identified critical UI/UX bottlenecks and potential error-forcing contexts within the software. The results demonstrate how a synergistic evaluation between AI and human expertise can enhance the robustness of safety-related software, ensuring that HRA practitioners can perform reliable risk assessments with minimized interface-induced biases.