Tutorial Overview

This guide shows how to build an optimization model with Modeler, the modeling agent in the Gurobi Intelligence Hub, using worked business problems.

This guide covers:

• Mathematical optimization — what mathematical optimization is, what it is used for, what a model is made of, and which part of the modeling job Modeler takes on. Start here if optimization is new to you.

• The methodology — the workflow behind every example: how the problem definition is built through questioning, why the specification is the central artifact, how generated code is validated by tests, how an optional mathematical formulation can be derived and verified, and why the workflow is a loop that keeps all of them in sync as the problem changes.

• Develop with Modeler — worked business problems, each applying that workflow end to end, from the first question through to a changed requirement.

• Advanced topics — four standalone topics: why use Modeler over steering a general-purpose LLM yourself, bringing in an existing model, Explainer’s diagnosis/repair/sensitivity workflow, and calling Modeler as an MCP tool.

This guide focuses on Modeler

The methodology and every page in Develop with Modeler are about Modeler. Diagnosed with Explainer works through the three Explainer use cases on an infeasible version of the facility model. Gurobot is not covered here.

If you already build models with an LLM coding assistant, Why not just an LLM coding assistant? covers what changes in the modeling work specifically.

For how to reach the agents from the portal or through the MCP server, the maturity of each one, and the model-size limits that apply to cloud solves, see the User Manual.