Honest Review: v0 by Vercel
We used v0 on three real projects. It's impressive for prototyping, but production is a different story.
v0 has been the most hyped AI design tool since it launched. Every week there's a new viral tweet showing someone build a complete UI in minutes.
We used it on three real client projects. Here's what happened.
What v0 does well
The initial generation quality is genuinely good. Describe a dashboard layout, a settings page, or a marketing section, and v0 produces something that looks professional out of the box.
The shadcn/ui integration is seamless. Components come out using the same primitives we'd choose manually. The Tailwind output is clean — not the utility-class soup you get from some generators.
For client presentations and early concepts, v0 is exceptional. We've cut our mockup time by half on projects where the client needs to see something visual before committing.
Where it falls apart
The moment you need to integrate v0 output into an existing codebase, friction appears.
Component structure rarely matches your project's patterns. v0 makes its own decisions about state management, data flow, and component boundaries. Those decisions are reasonable in isolation but wrong for your specific context.
Responsive behavior needs manual work. The generated layouts look good at the viewport size you're previewing, but breakpoint behavior is often simplified. Real responsive design requires understanding the content priorities at each size.
Accessibility is surface-level. ARIA labels exist, but keyboard navigation patterns, focus management, and screen reader announcements usually need work.
Project 1: Marketing site
Used v0 to generate hero sections, feature grids, and pricing tables. Kept about 60% of the output. The visual quality was high, but we restructured every component to match our patterns.
Verdict: Good accelerator. Saved maybe 4 hours on a 2-day project.
Project 2: Dashboard interface
Generated a data table with filters, a sidebar nav, and a settings form. Kept about 30% of the output. The dashboard layout was too generic — it didn't account for the specific data relationships our app needed.
Verdict: Better starting point than blank file, but barely. The customization time nearly matched writing from scratch.
Project 3: Prototype for investor demo
Built an entire 5-page prototype in v0 for a client pitch. Kept 80%+ because production quality wasn't the goal — visual communication was.
Verdict: This is v0's sweet spot. Rapid visual prototyping where code quality is secondary.
The real assessment
v0 is a prototyping tool that produces production-adjacent code. If you treat it that way, it's excellent. If you expect production-ready components, you'll be disappointed.
Use it for concepts. Use it for client presentations. Use it to explore layout ideas quickly. Then rebuild the parts that matter with your own engineering standards.
That's not a criticism. That's understanding the tool for what it actually is.