Building, optimizing and evaluating an agentic workflow with guardrails, fine-tuning, and extensive evaluation. We also use this case study to explore responsible and fair agent development and reporting for the EU AI Act.