Building a custom analytics dashboard usually means days of boilerplate: app scaffolding, callbacks, layout wiring, and database plumbing. All of that happens before you can even ask whether the dashboard is answering the right questions. This post walks through a small experiment: how far I could get building an Ironman training dashboard with Google Antigravity using minimal prompting, what that revealed about where LLMs accelerate dashboard development and where human judgment still matters most. The goal wasn’t a production-ready app. It was to shorten the distance between idea and working prototype.