International Journal of Scientific Research and Engineering Development

Most AI writing tools generate documents in a single pass, forcing users to spend hours manually editing the output. While newer systems attempt self-improvement through internal feedback, they ignore what users actually want to change. This paper proposes FeedbackRefine, a conceptual framework that lets users guide document improvement through iterative feedback cycles. The key idea is simple: generate an initial draft, ask the user for feedback, use that feedback to improve the document, and repeat until the quality is acceptable. FeedbackRefine operates in five stages: (1) initial draft generation, (2) quality evaluation against a multi-dimensional rubric, (3) user feedback collection, (4) feedback transformation into refinement instructions, and (5) targeted document refinement. This framework addresses a clear gap in current research: existing systems either produce one-shot documents or rely on internal AI feedback, but rarely incorporate actual user input to guide refinement. We propose evaluation methods to test whether user-guided iteration reduces editing time and improves document quality compared to standard approaches. This work contributes a conceptual foundation for building better human-AI document creation systems and identifies key research questions for future empirical validation.

C. Sreerag, Prof. Dr. Shivani Budhkar,"FeedbackRefine: User-Guided Refinement of LLM-Generated Professional Documents" International Journal of Scientific Research and Engineering Development, V9(3): Page(1235-1238) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.