I Can Build That With AI Trap

With Great Power Comes Great Responsibility.

The first time I saw that quote was the first time I opened a terminal on Linux. As a kid, I’d always admired Hollywood’s depiction of hackers — the rapid-fire typing, the dramatic countdown, the triumphant shout when they finally brought down the evil corporation. That first encounter with a terminal gave me a sense of purpose. I had an opportunity to learn, to grow, and to take my place among the programmers I admired. Since then, my interest in technology has taken me through different programming languages, databases, servers, and industries.

Today, we’re in a brave new world of AI. You no longer need a PhD to achieve remarkable things with it. With the rise of vibe coding, almost anyone with a spark of an idea believes they can leap straight to building a software product for their dream.

The industry has embraced this shift across almost every domain. I mean, I can barely tell whether a YouTube ad features real humans or AI-generated actors anymore — these tools are getting frighteningly good.

The Confidence Trap

With so many AI-powered building platforms flooding the market, it’s now possible for anyone to build whatever software they want — and if the idea is good enough, turn it into a business. There are countless success stories and just as many horror stories about vibe-coded products, but let’s keep going.

The real threat to SaaS businesses is that people now think: why pay for a subscription when I can vibe code the same thing for $20 a month on an AI platform?

I fell into this trap myself. Whenever I publish a blog post, I schedule and distribute it across my social networks using Buffer. I’ve been on the free tier for years, well aware I’m missing out on their analytics features — but I never thought I’d care. I just wanted to schedule posts.

Then I noticed my blog traffic increasing over the last ninety days, and I got curious: were people actually visiting through the links I post on LinkedIn? It doesn’t help that LinkedIn overrides and wraps your links in their own tracker. Despite that, it seems I do get a fair number of direct referrals. Am I ready to pay for Buffer’s premium plan? Nope.

So naturally, I thought: what if I just built it? A frontend to write posts, a simple database to queue them, social media integrations, and URL tracking to measure blog traffic. It all sounds easy enough — the kind of thing I could probably get AI to scaffold in an afternoon. Then the next question hit me: what about operations?

The Operations Trap

Now you’ve vibe coded a Buffer clone. Where do you host it?

Hopefully you chose SQLite and not one of the larger managed databases, because that means another subscription. You’ll need a deployment pipeline too — because you’ll inevitably want to improve the thing you just built. And now this hobby project becomes yet another application on your ongoing maintenance schedule.

I’m an experienced software engineer, and I still felt a wave of exhaustion just thinking about it. Now imagine someone who discovered software development through vibe coding last month. Can AI guide you through deployment? Yes — if you already have some experience. If you don’t, welcome to hell.

Build What’s Unique. Pay for the Rest.

Before you go down the rabbit hole, ask yourself:

• Am I willing to take on ongoing maintenance and operations?
• What do I actually gain from cloning this service — cost reduction, or can I genuinely do a better job?
• Is this just fun with AI?

Based on your answers, you might decide to build the clone anyway, or just pay for the service and move on.

As for me? I’m still going to build the Buffer clone — out of curiosity — and while I’m at it, rescue my bookmarks trapped in Instapaper. Some lessons you have to learn for yourself.