A Tale of Two Weekend Projects

When I was in college, my friends and I would come up with a fun name for a product and cobble together some code copied from tutorials and Stack Overflow pages. Then we’d buy a domain name, publish the site, and proceed to get banned from subreddits for self promotion (making no money in the process).

This was our “vibe coding”.

When “Relaweather”, our web app for reporting today’s weather based on a comparison to yesterday’s, didn’t take off, we weren’t too concerned. We had a bit of fun, learned some things, and called it a day.

Today, a budding entrepreneur using vibe coding tools can create a fully functioning prototype of Relaweather.ai (“the Agentic Meteorology platform for the AI-native era”) in a weekend. They’ll have a cool app launched on the App Store and a slick site with a pricing page offering a $9/month subscription ($99 annually, save 10%).

While 2026’s Relaweather.ai prototype might look a lot nicer than the one we made in 2014, they both have essentially no economic value.

If you’re interested in making real software, get ready to put in real effort. Get ready to make real buttons and store real data in a real database. Once that’s done, if you’re lucky, customers might pay you real money.

Oh wait, that’s not the end?

What the hell??

The password reset button doesn’t work? The lightning animation doesn’t work on Firefox? I thought I fixed that yesterday. Ah damn, the deploy failed! Well, not a problem, we’ll just run the build again… etc. etc.

This is software.

It takes more than a weekend, but if you stick with it you’ll create real value. Eventually, those customers will thank you, and recommend your thing to their friends, or someone will post about you on Reddit and it won’t be self promotion.

Don’t let your AI prototype trick you into thinking you’re done after a weekend.

You’re not.