Synthetic Ecosystems: Using Generative Agents to Simulate Urban Policy Outcomes in Virtual Cityscapes

Abstract

This study introduces a novel simulation framework that leverages generative agents to model synthetic urban populations and predict the impacts of various policy interventions. By fusing real-world demographic, economic, and environmental datasets within digital twin representations of city environments, the platform enables policymakers to test and evaluate urban policy scenarios in silico. The framework integrates a generative agent engine with dynamic data ingestion pipelines, allowing for adaptive behavior modeling across social, economic, and ecological domains. Applied to the Amsterdam Smart District, the model simulates policy scenarios such as housing reforms, universal basic income (UBI) implementation, and smart mobility initiatives. Results are visualized through predictive dashboards that highlight key urban indicators and comparative outcomes. The system demonstrates promising accuracy in forecasting policy impacts and offers potential for real-time integration with municipal open data platforms. This work presents a step toward intelligent urban governance by enabling evidence-based decision-making in virtual cityscapes.