While Large Language Models (LLMs) excel at single-document queries and conversational workflows, they struggle with progressively exploring, analyzing, and synthesizing large unstructured document sets, such as in literature surveys. We address this …