Using LLM Agents for Strategic Goals Extraction and Decision-Making

Abstract

Strategic planning is vital for organizations, but drafting institutional strategies and aligned unit-level goals is complex. Large Language Models (LLMs) can support this through language understanding and generation, though direct use risks unsupported, misaligned, or overly automated decisions. This project presents StratAI, a web application developed as a supervised LLM-agentbased decision-support system for institutional strategy drafting and administrative-unitlevel strategic goal derivation. The system is designed around an architecture in which LLM-based capabilities are used inside controlled stages rather than as fully autonomous decision-makers