The experiment
Every prior discussion of the app’s first-run experience had been secondhand
— someone describing what the editor does, not a picture of what it actually
looks like on load. I wanted primary evidence, not a paraphrase, so I had
Playwright drive the real app and capture real screenshots: landing on
app.buildmy.house cold, opening the editor, placing a wall, opening the
properties panel, and hitting save.
What the screenshots showed
Four problems, all in that same first-run path, none of which had been reported before because no one had looked at the actual pixels:
- Menu bar CSS bug — the top menu bar’s items overlapped at the viewport width Playwright defaulted to, legible but visibly broken.
- Empty 3D backdrop — the editor’s viewport loads as a flat void with no ground plane or horizon reference until something is placed in it, so a new visitor’s very first frame of “the product” is emptiness.
- “Sign Up to Save” prompt — it fires the moment you touch the save action, before there’s anything meaningfully designed yet to lose, directly undercutting the “no signup” claim the homepage is about to lead with.
- Empty properties panel — before an object is selected, the panel renders as a blank box with no placeholder state, reading as broken rather than simply idle.
Why this mattered beyond the bug list
This is the exact path the new homepage positioning (“no download, no signup”) is about to send a lot more traffic through. Every one of these four issues sits directly in that path, and each one contradicts the claim we’re about to make louder. Finding them with screenshots instead of descriptions is what made them undeniable enough to actually fix rather than get triaged into someday: a menu bar CSS fix, a default backdrop scene, moving the save prompt to a point where there’s something worth saving, and a real empty-state for the properties panel. All four are now fixed, and the screenshots are saved as the baseline to catch regressions the next time this path changes.