Optimization remarks are a powerful yet underutilized tool for gaining insight into compiler heuristics and transformations. Remarks can inform the user on which optimizations applied successfully, which failed to apply, and which heuristics led to those decisions. We want to provide users with actionable, low-noise remarks about possible performance oversights in their code, and compiler developers with fine-grained per-function telemetry to qualify optimizer changes on large codebases and track down regressions. Our goal is to provide a unified remarks framework to cover both use-cases. In this talk, we will provide an overview of the state of optimization remarks, how to use them, and how to further extend them. We will show examples for applying remarks to different use-cases (e.g. improved vectorization failure hints, inlining summaries, regression analysis).