ModelTriage

ModelTriage is a web app (with optional desktop agent) for teams building complex system models—agent-based, system dynamics, or hybrid simulations—who are drowning in parameters and uncertain assumptions. You upload a model package (NetLogo, Python, Julia, or CSV-based equations) and define outputs you care about. The app runs structured sensitivity analysis and uncertainty quantification to identify which parameters drive outcomes, where tipping behavior appears, and which assumptions are effectively irrelevant. It generates a ranked “effort-to-impact” list: what to measure next, what to calibrate, and what to ignore. It also produces audit-ready reports with reproducible run configs and versioned datasets so teams can defend decisions to stakeholders. This is not a magic oracle: it won’t validate a bad model, but it will quickly expose fragility and reduce wasted experimentation time.