Marine ecosystem models are among the most powerful tools available for anticipating how oceans will respond to change and informing management decisions. But the technical demands of developing, parameterising, and running these models mean that their use depends on a small pool of specialist modellers, a bottleneck that is increasingly untenable as communities confront rapid ecological and social change. The people with the most intimate knowledge of local systems, and the most urgent need for decision-support, often cannot access modelling capacity when they need it.
Here I present a desktop application that combines an interactive modelling environment with AI-assisted guidance to make ecosystem model construction more accessible, transparent, and reproducible. Users define study regions, upload or generate spatial layers, select active functional groups, and parameterise ecological relationships through a graphical interface. An integrated large language model copilot assists through each stage of the model-building workflow, helping translate ecological knowledge into model parameters while maintaining full provenance tracking of data sources and parameter justifications.
The platform is designed to support multiple modelling backends; I demonstrate it here using CORSA, a spatially resolved food-web framework with over 29 ecological modules spanning primary producers, consumers, biogeochemical cycling, habitat dynamics, and anthropogenic pressures. I illustrate the workflow through an application to Australian shelf ecosystems, showing how AI-mediated model building can lower the technical threshold for participation while preserving scientific rigour.