Building Passport: A Full-Stack App in One Weekend

The Idea

I work in aviation analytics. After years of data work in the airline industry, I wanted something nobody had built yet: a way to visualize every flight I've ever taken on an interactive 3D globe.

Most people have flight maps. Mine plots 230 flights across 28 countries and 50 airports on a globe you can rotate, zoom, and explore. It's called Passport.

The Old Timeline

Six months ago, I would've started this project with a realistic estimate: six to eight weeks. That's what it takes to go from idea to a live web app when you're building it yourself.

Here's the stack that takes time:

• Frontend framework (React, Vue, or similar)
• 3D visualization library (Cesium, Three.js, Babylon)
• Backend API to serve flight data
• Database setup
• Deployment pipeline
• CSS that doesn't look broken

Each piece has a learning curve. Each integration point is a place to debug.

What Changed: Claude Code

One weekend. Start to finish.

Here's how it actually happened. I had a clear idea of what I wanted to build. I knew the tech stack that would work. I had the domain knowledge (I know aviation and flight data). What I didn't have: the time to write all that boilerplate.

That's where Claude Code made the difference.

Instead of writing the code myself, I described what I wanted. Claude wrote the skeleton, the APIs, the components. I focused on the logic that actually mattered: getting the flight data right, making sure the globe worked, tweaking the visual design until it felt right.

We shipped Passport. Live. Usable. Working.

The Real Lesson

What made this work wasn't just Claude. It was the combination. I brought something Claude couldn't: I knew aviation. I could spot when the flight data was wrong. I understood what a good visualization should actually show. I had taste.

Claude brought speed. It wrote the boilerplate. It handled the integration points. It knew the libraries and frameworks I didn't need to waste time learning.

What happened between us was the iteration. I didn't use Claude's first draft. I tweaked it, broke it, asked for something different, refined it. That cycle is what turned a good idea into something actually worth shipping.

The mistake people make is thinking AI replaces thinking. It doesn't. It just removes the friction around the things that aren't thinking. I didn't spend time writing React boilerplate. I spent time understanding the problem.

If you've got domain expertise, a clear vision, and the willingness to iterate without being precious about the first version, AI coding works. If you're vague about what you want or you don't know if something is right when you see it, no amount of AI will help.

The same thing applies to your work. The competitive advantage won't go to the company with the fanciest tech stack or the person who knows AI the best. It goes to whoever understands their problem deeply enough to know what to build, then has the speed to actually build it.

Passport started as a weekend project. It's a reminder that the bottleneck in building isn't the coding anymore. It's the thinking. It's knowing what to build. That's where your time should actually go.

Try it: https://lnkd.in/eHrKZnJD

What's one thing you've been putting off building?