Jairath et al. (2024) introduced ReconGPT, which is a GPT-4-basedRetrieval-Augmented Generation (RAG) model with visual capabilities designed to assist dermatologic surgeons in decision-making regarding post-Mohs surgical reconstructions. It provides this assistance in the form of appropriate reconstructive techniques or possible surgical complications for given anatomic defect locations. Using ReconGPT, over 88 clinical images were evaluated with a different reconstructive challenge. This model was meant to identify and predict potential repair strategies for surgical complications. ReconGPT achieved 97.7% concordance in defect localization, 60.2% in general repair predictions, 30.7% for specific repairs, and 47.6% for complication prediction. ReconGPT was often stronger for single-modality repairs and weaker for multi-modality strategies or specific anatomic regions like the eyelid or nose. The authors note that ReconGPT cannot replicate surgeon intuition or account for patient preferences; however, this software has potential as an adjunct to surgical decision making.