Generative AI for NSF Grant Prewriting: Zero-Shot Protocol Prompting for Tactical Competitive Intelligence

Abstract

Libraries are increasingly positioned to support faculty grant development, yet proposal-prewriting assistance remains uneven across institutions, particularly where research-development infrastructure is limited. This article presents a practical generative AI workflow for transforming publicly available National Science Foundation award abstracts into tactical competitive intelligence for early-stage grant planning. We call this approach zero-shot protocol prompting: a method in which users upload a structured protocol file and an NSF award-abstract corpus into a generative AI system, then issue a brief instruction that produces coded outputs and a faculty-facing synthesis document. Drawing on writing-process theory, proposal-development literature, competitive intelligence, prompt design, and library AI ethics, the article explains how the workflow was developed, tested, and refined through human-AI prompt iteration. The resulting process supports applicants in identifying patterns in previously funded projects, including common methods, broader impacts language, intellectual merit framing, disciplinary trends, and evidence of sponsor alignment. Rather than using AI to draft proposal text, the workflow preserves the human writer as the primary author while using AI to organize public information for reflection, consultation, and ideation. The article argues that zero-shot protocol prompting offers libraries a low-cost, scalable, and ethically grounded way to expand proposal-development support, especially at institutions with limited staffing or formal grant-writing infrastructure.