← ShroomifyPLAYGROUND

The side project I used to stress-test AI coders.

ROLE

Hacker

WHEN

2025

STATUS

Concept

#Claude#V0

The benchmark

I ran the same brief through Claude and V0 and compared what came back. And it was a real brief, the kind an actual design team would hand over: branding direction, app architecture, a user flow, and lo-fi Figma screens.

The lo-fi Figma screens from the brief: sign-in, ordering a kit, the space check, unpacking, and harvest, with Planty the AI assistant guiding each step

The short version: Claude leaned into engagement and interaction — clarifying questions, a plan before it built, hover states on everything; V0 jumped straight to building with denser, more "website" content blocks. Claude = engagement + interaction; V0 = content + structure. The point was never a winner; it was learning where each tool's instincts land.

V0's build: the Shroomify home screen, glowing mushroom header and all

Claude's build after a recovery run: the sign-up screen in the brief's fuchsia-and-cyan branding

→ Full write-up: Claude vs V0: testing AI vibe coding for Shroomify.

Field notes

Shroomify did ship, as an app prototype rather than a finished product. You can try it live (source) and click through the grow flow yourself. What I keep from this project, though, is the harness: a real brief, handed to each tool, that showed exactly where it added value and where it broke.