The fastest way to end a two-week argument about what a product should do is to put a working version in front of the people arguing. Opinions about a slide never end. Reactions to real software are specific: this button is in the wrong place, this number is wrong, this is the screen I'd actually use.
Building a first version is now cheap enough that it's often quicker than writing the full spec. The prototype becomes the spec, and the conversation moves from what we might build to what we should change.
So where we can show it working, we do.