Two directions, one very useful distinction
Composition-to-property prediction and property-to-composition design solve different problems. Having both in the same workflow is appealing. A side-by-side example using the same material system would explain the difference faster than a long feature list.
Great framing — forward and inverse design really are complementary directions of the same problem, and the value shows up when they share a workflow.
We agree that a side-by-side demo is the clearest way to show this: take one material system, predict its properties from a given composition, then run the inverse direction from the target properties and see where the candidates land. We're putting together exactly that example for the docs.
If there's a material system you'd find most illustrative (e.g., a specific alloy family), let us know — we'd like the demo to reflect real use cases.