Vibe Coding an LLM-based demo project: Life Clock

Based on the course instructions and provided demo files, I built the Life Clock as another learning project.

Setting up the app and API connection in VS Code

Life Clock running locally — Life Clock working locally

Here is how the LLM explains what we built together, in plain language.

What we built

We have a small web application running on your Mac. Here’s what each piece does.

The Server — app.py

This is the engine running in the background. When you type a food item and hit ✨, your browser sends that text to the server. The server takes your text, bundles it together with the RAG file, and sends both to Claude over the internet. Claude sends back a response, the server unpacks it, and sends it to your browser to display.

Think of it like a waiter — it takes your order, goes to the kitchen (Claude), and brings back the food.

The RAG File — life_copilot_rag_registry_cowboy.txt

This is the instruction manual the server hands to Claude with every single request. It tells Claude to behave like a Texas cowboy, how to score food choices, what attributes to assign, and how to format the response. Claude reads it fresh every time you submit something.

Think of it as the chef’s recipe book — the kitchen follows it every time, no matter who orders.

The .env File

A small hidden file that holds your Anthropic API key — the password that lets your server talk to Claude. It never leaves your machine.

The Template — index.html

This is purely what you see in the browser. It has no intelligence of its own — it just displays whatever the server hands it. The Quicksand font, the green buttons, the pinned input bar — all of that lives here.

JSON

The format we asked Claude to respond in. Instead of one blob of text, Claude sends back a structured package with separate fields — the cowboy message, the attributes, the lifespan impact. This lets the template display each piece independently, like placing ingredients on a plate rather than dumping everything in a bowl.

The Flow in One Sentence

You type something → server adds the RAG instructions → Claude responds in JSON → server unpacks it → your browser displays the pieces in the right places.

And just for fun, here is a bullsh*tty LinkedIn version:

🚀 Thrilled to announce that I have single-handedly disrupted the $4.2 trillion global wellness industry.

After months of deep strategic vision and relentless execution, I have shipped Life Clock — an AI-native, RAG-augmented, full-stack health intelligence platform with a conversational interface paradigm and real-time biometric attribute scoring engine.

The architecture? A proprietary Python microservice with a dynamic prompt orchestration layer, JSON-structured LLM response normalization, and a persona-driven conversational UX framework informed by Texan frontier philosophy.

The design system? Pixel-perfect. The color palette? Intentional. The font? Quicksand. You wouldn’t understand.

We’re not just tracking what you eat. We’re quantifying your life. In minutes. The waitlist is closed. The vision is open.

#AI #HealthTech #Disruption #FounderLife #RAG #LLM #FullStack #WellnessTech #TheFutureIsNow #Quicksand 🤠✨