The idea started with a frustration: every explainer I found about “the AI supply chain” assumed you already understood half the stack. Power, chips, data centers, models, applications — each article picked one layer and went deep, but nobody connected them into a single picture a beginner could hold in their head. So I decided to build the explainer I wished existed.
I didn’t write a line of code myself. What I brought was the brief: what the tool should teach, who it was for, what “done” looked like. Claude Code brought the implementation. That division of labor is the whole idea behind vibe-coding — you’re not abdicating judgment, you’re redirecting it. Instead of deciding where a semicolon goes, you’re deciding whether the third section actually explains the idea or just restates it with a fancier verb.
“A good brief is worth more than a week of trial and error.”
The weekend, hour by hour
Saturday morning was entirely spec. I described the six layers I wanted to cover — energy, hardware, cloud infrastructure, model training, model serving, and the applications people actually touch — and Claude Code proposed an interactive scroll-based layout to walk through them in order. We iterated on the structure twice before any real code existed, which felt slow in the moment and saved hours later.
Layers:
1. Power & Cooling
2. Chips & Hardware
3. Cloud Infrastructure
4. Model Training
5. Model Serving
6. Applications (what you actually see)
By Saturday evening, the scaffold was live on localhost, and Sunday was almost entirely refinement — copy passes, mobile layout fixes, and one very satisfying hour making the scroll transitions feel less like a slideshow and more like a story.
It shipped Sunday night. No prior coding background, no bootcamp, no computer science degree — just a clear brief, a lot of back-and-forth refinement, and a tool that could turn both into working software. That’s the proof I keep pointing to when people ask whether this approach actually scales beyond a toy project: it’s live, it’s real, and people use it.