PolicySwarm

PolicySwarm is a web app (with optional desktop runner) for building and running agent-based models focused on public policy and operations decisions: housing, transit, public health, campus safety, and retail foot-traffic. Users assemble agents, environments, and rules from a constrained library (to avoid “anything goes” models), import real datasets (census, mobility, POS, staffing), and run scenario sweeps with clear uncertainty ranges. The app outputs decision-ready artifacts: sensitivity charts, distributional impact by subgroup, and “assumption cards” that make hidden choices explicit. It’s a combination traditional + AI app: AI helps translate plain-language policy questions into model scaffolds, suggests parameter priors from referenced sources, and flags non-identifiable/overfit dynamics. It won’t replace expert modelers; it aims to make ABM usable for teams that currently avoid it because tools are too academic or too bespoke.