STORM

STORM (Synthesis of Topic Outlines through Retrieval and Multi-perspective Question Asking) is an LLM system that automates the pre-writing stage of generating grounded, long-form articles.

From STORM-paper

• STORM models the pre-writing stage by discovering diverse perspectives, simulating conversations with topic experts, and curating information into an outline.
• It uses perspective-guided question asking to improve the depth and breadth of research.
• The system is implemented using the DSPy framework.
• It significantly outperforms baseline RAG methods in article organization and coverage.