Adaptive Iterative Elicitation: A Prompt Pattern for Personalized LLM Decision Support

Large Language Models (LLMs) have demonstrated remarkable capabilities in generating re- sponses to user queries. However, their default question-answering paradigm often produces generic recommendations that fail to account for individual context, preferences, and constraints. We present the Adaptive Iterative Elicitation (AIE) prompt pattern—a systematized framework that trans- forms LLMs from passive responders into active conversational advisors. The core principle is sim- ple yet powerful: ask one focused question at a time, use each response to guide the next question, and continue until sufficient clarity is achieved to provide personalized recommendations. Unlike existing Socratic dialogue systems that focus primarily on education, AIE is designed to generalize across diverse decision-support domains including career guidance, health and wellness, financial planning, software architecture, and general concierge services. We present the framework’s de- sign principles, illustrate its application through an exploratory career counseling deployment, and provide instantiations across multiple domains. Preliminary observations suggest the pattern en- ables more personalized outcomes compared to traditional single-prompt interactions. We release the framework and prompt templates to enable broader adoption and invite the research community to conduct rigorous empirical evaluations.