The profitability of cloud providers is often negatively affected by misconfiguration of application resource constraints. In this research study, we check the feasibility of integrating ML on usage-dependent configurations into a GitOps workflow. The result will be a novel advisor service that tells GitOps engineers about monetary implications of detected misconfigurations. Associated research questions are: Can existing ML components be integrated into GitOps workflows? Are they scalable enough to process metrics from a growing customer base? Do they identify the parts of the configuration that are the low-hanging fruits in the sense of saving most resource expenses after adopting few suggested changes?