Democratizing AI Social Simulation: A No-Code Web Interface for the Concordia Framework

Generative agent simulations offer transformative potential for computational social science, yet the requirement for sophisticated programming expertise limits their adoption to technical specialists, excluding many expert domains. This paper presents an open-source, web-based platform that democratizes access to the Concordia framework through an intuitive interface, allowing researchers to configure, execute, and analyze multi-agent simulations without writing code. Key technical contributions include a reliability layer for sustaining long-running simulations with automatic checkpointing, mid-simulation cancellation, and timeout protection to mitigate LLM API instability, and a provider-agnostic adapter supporting both commercial and locally hosted models. The platform streamlines research workflows through 38 scenario templates spanning 8 game master types, census-based agent generation from demographic distributions, batch execution with parameter sweeps, structured data export, and automated LLM-powered analysis. We evaluate the system across diverse scenarios and discuss trade-offs between this interface and the native Concordia Python framework. By eliminating programming barriers while preserving advanced simulation capabilities, this tool serves as both an accessible entry point for non-technical researchers and a rapid prototyping environment for experts. This is the accepted manuscript of a paper presented at SIMULTECH 2026. The final version will be published in the proceedings of the conference by SCITEPRESS.