How I Built a YouTube Analytics Tool in One Evening (With AI)

Last week I got a challenge from a startup: build a YouTube competitor analytics tool from scratch.

No specs. No design files. Just a rough idea and a deadline.

Here's how it went.

The idea was simple. Paste any YouTube channel URL → instantly see which videos are performing best right now.

Not just raw views. Actual performance signals.

What I built:

• Velocity-based trending detection (views/day vs channel average)
• Top Performers section — top 3 videos highlighted instantly
• Quick Insights — auto-generated summaries from computed stats
• Bar chart, sortable table, grid/list toggle
• CSV export
• Full mobile responsiveness

The stack: Next.js 14 · Tailwind CSS · Recharts · YouTube Data API v3 · Vercel

The process: I used an AI-native workflow throughout.

Claude Code handled execution — component generation, API integration, UI iteration. Abacus ChatLLM handled product thinking — feature prioritization, UX decisions, iteration strategy.

I treated AI as a collaborator, not a shortcut. Every product decision — what to build, what to skip, what to prioritize — was mine.

The result: A deployed, demo-ready product. In one evening.

👉 Live: https://tubepulse-one.vercel.app/ 👉 GitHub: https://github.com/Valdas2020/tubepulse/

What I learned:

The bottleneck in modern product development is no longer writing code. It's knowing what to build and in what order.

AI handles the execution layer. You handle the judgment layer.

That's the whole game now. Try it yourself → https://tubepulse-one.vercel.app/

Built with Claude Code + Abacus ChatLLM. Stack: Next.js · Tailwind · YouTube API · Vercel

#vibecoding #buildinpublic #AI #nextjs #productdevelopment #youtube