AI agents can already handle parts of the modeling workflow that once took hours by hand: writing Mlxtran code, running PK workflows, generating diagnostics. That shift makes one skill more valuable than ever — the scientific judgment to know whether the output is right.

This two-day workshop is built around that idea in three parts. Through lectures and hands-on exercises in MonolixSuite, you will build both fluency with the software and the deeper understanding that lets you work critically, whether you are using the interface yourself or supervising an AI agent.

• Part 1: Modeling We cover the full modeling workflow. You will explore a dataset, build structural and statistical models, run a covariate search, and develop a PK-PD model. Alongside the practical work, we make time for concepts often skipped: what the estimation algorithms are actually doing, and how to read diagnostic plots with real understanding.
• Part 2: Simulations We start with the question and then the practical skills to build simulations: which types of simulation you can run, which sources of uncertainty to include and why, and how to analyse the results, including testing differences statistically rather than by eye.
• Part 3: AI We open the black box: what a large language model (LLM) is, how to build an agent, and how prompting, context and tool access shape what an AI system produces. From there we look at what makes a good modeling agent, how to actually do modeling with one, and where your judgment and review stay essential.

This workshop does not start from zero. You should be able to run a basic PK workflow in Monolix: load a dataset, set up a model, run an estimation. If you are not there yet, we offer a free online preparatory course covering the essentials, so you arrive ready to benefit fully from both days.

Includes a temporary license for the MonolixSuite.