A smaller first release.

Keep the learning intact

I would scope a first release around one complete user outcome. That may mean fewer inputs, a narrower audience, or a manual review at an important transition. The useful constraint is that the path still reaches something someone can use and evaluate.

A small release should answer a question about the design. Can people understand the result? Does the handoff work? Is the difficult part actually difficult in practice? I want the next scope decision to come from those observations. That keeps a reduced first version meaningful while leaving room to change the larger plan.