Maybe the language/framework/GUI could generate or require unit/integration/system tests (with just a few clicks to confirm or correct an assumption), and visua

Maybe the language/framework/GUI could generate or require unit/integration/system tests (with just a few clicks to confirm or correct an assumption), and visualize in broad strokes which scenarios or input ranges are "green" yet. That would mean a lot more infrastructure (sorry if that wasn't what you meant), but automatically provided or accepted by everyone involved.