You’re coding late at night. The lights are low. Your favorite playlist is playing, maybe something only you would call “focus music.” Your IDE is set to a theme that feels just right. You’re not rushing. You’re not distracted. You’re in the zone.

This is vibe coding.

But it’s not just about music and mood. Vibe coding also means working with AI tools that support your thinking. Vibe coding was introduced by Andrej Karpathy and works like this: you sketch the idea in plain language, the AI writes the code, you run it, watch what it does, and accept all changes without reviewing or editing any code.

“fully give in to the vibes, embrace exponentials, and forget that the code even exists.”

In this post, I will focus on that second part: coding with AI support. Because while the right playlist might help you get into flow, you can’t escape from the AI hype.

What vibe coding actually looks like, let’s make it concrete.

You’re building a backend service in Java. You start by asking Claude to outline a repository pattern. Ask Copilot to fill in the repetitive parts. You’re thinking, typing, and building in one continuous motion, fully integrated in your IDE. No need to dive into a forum to find some specific detail or open the Design Patterns book of The Gang of Four.

The music helps of course but it’s AI that keeps you in the vibe.

Why does it work

Vibe coding works because it aligns with how we naturally want to work. It reduces the mental overhead. It removes disruptions and the small blockers. It creates space to think.

What does it offer

  • Faster ideation:

    • You can move from idea to implementation without losing momentum.

  • Less context switching:

    • You stay in your editor, in your flow, without jumping between tabs.

  • Better focus:

    • The combination of ambiance and assistance helps you stay in the zone.

It’s no silver bullet

Vibe coding feels great, but it’s not without its risks.

  • Over-reliance on AI:

    • You might stop thinking critically if the AI always has an answer.

  • Loss of craftsmanship:

    • Aesthetic-driven choices overshadow security, code quality, clarity, and maintainability

  • Loss of depth:

    • If you let the AI write everything, you miss the chance to learn.

  • Echo chamber effect:

    • AI can reinforce your own biases if you’re not careful.

Like all my blogs, I love to add some Michael Jordan wisdom:

vibecodingMJ

The minute you get away from fundamentals, whether it’s proper technique, work ethic or mental preparation, the bottom can fall out of your game, your schoolwork, your job, whatever you’re doing.

— Michael Jordan

This quote resonates strongly with coding: when you lose sight of clean architecture, thoughtful design, or disciplined habits, the vibe might still feel good, but the foundation starts to crack.

So… hype or vibe?

Vibe coding isn’t a methodology. It’s not a framework. It’s a way of working that can support you if you use it the right way. It’s a way that can make coding more enjoyable, and sometimes even more productive.

Vibe coding works for quick prototypes and rapid application development, it’s great for exploring ideas and getting to “something that works” fast. This can be categorized under the emerging term Disposable Code.

But for security sensitive applications and codebases that we as human developers must maintain, it’s not so straightforward. You’ll need strong reviews, tests, documentation, and clear ownership.

Let the AI assist you, not replace you. Let the vibe support you, not distract you. Keep your standards high, even when the mood is low.

It only works if you’re in charge!

shadow-left