City-LEO: Toward Transparent City Management Using LLM with End-to-End Optimization

TL;DR

City-LEO addresses the transparency and complexity barriers of traditional OR tools in smart-city operations by integrating an LLM-based agent with an End-to-End optimization framework. It uses LLM-driven problem scoping to reduce large-scale optimization problems to neighborhood-level decisions, enabling faster, more relevant solutions while maintaining core operational objectives through a Random Forest–based E2E model encoded as a MIP. The approach is validated on a Seattle e-bike sharing case, showing improved query relevance and substantial computational gains over full-scale optimization, with acceptable trade-offs in optimality. The work demonstrates the potential of LLM-embedded OR tools to enhance transparency, responsiveness, and public trust in urban governance, and points to future work on real-time decision-making and feedback integration.

Abstract

Existing operations research (OR) models and tools play indispensable roles in smart-city operations, yet their practical implementation is limited by the complexity of modeling and deficiencies in optimization proficiency. To generate more relevant and accurate solutions to users' requirements, we propose a large language model (LLM)-based agent ("City-LEO") that enhances the efficiency and transparency of city management through conversational interactions. Specifically, to accommodate diverse users' requirements and enhance computational tractability, City-LEO leverages LLM's logical reasoning capabilities on prior knowledge to scope down large-scale optimization problems efficiently. In the human-like decision process, City-LEO also incorporates End-to-end (E2E) model to synergize the prediction and optimization. The E2E framework be conducive to coping with environmental uncertainties and involving more query-relevant features, and then facilitates transparent and interpretable decision-making process. In case study, we employ City-LEO in the operations management of e-bike sharing (EBS) system. The numerical results demonstrate that City-LEO has superior performance when benchmarks against the full-scale optimization problem. With less computational time, City-LEO generates more satisfactory and relevant solutions to the users' requirements, and achieves lower global suboptimality without significantly compromising accuracy. In a broader sense, our proposed agent offers promise to develop LLM-embedded OR tools for smart-city operations management.

Figures (8)

• Figure 1: Dilemma between Relevance and Accuracy of LLM-based OR Tools in City Management
• Figure 2: Framework of the City-LEO Agent
• Figure 3: Prompt in Problem Matcher and QR-obj Generator
• Figure 4: Prompt in the Problem Tailor
• Figure 5: Sample Tree of the Trained Random Forest