A financial services firm wants to automate loan decisions using an ML model. Each record includes third-party credit history and customer demographics. Every prediction must include an explanation showing why the applicant was approved or denied. Using Amazon SageMaker, which solution provides this functionality with the least development effort?

SageMaker Clarify is the correct choice because it is a fully managed service within SageMaker designed specifically for bias detection and explainability. It can generate various explanation reports, including SHAP (SHapley Additive exPlanations) values, which are crucial for understanding feature contributions to individual predictions, directly addressing the requirement for explaining approval or denial reasons. This minimizes development effort compared to building custom solutions. SageMaker Model Debugger focuses on monitoring training jobs for issues like overfitting or underfitting, not on explaining individual model predictions post-training. AWS Lambda could be used to compute feature importances and partial-dependence plots, but this requires significant custom development and integration, which contradicts the "least development effort" requirement. Custom Amazon CloudWatch metrics are for monitoring operational performance and resource utilization, not for generating model explanations.