Not too long ago, I was scrolling through YouTube when I stumbled upon a video from Theo titled My Thoughts on Vibe Coding. At first, I had no clue what vibe coding was. I was like, What is this? What’s going on? And suddenly, it was everywhere—trending on X, YouTube, and the entire software development community.

If you’re curious, check out Theo’s video:

It’s a great watch, and I definitely learned a lot from it. But after watching, I had more questions than answers. So, I decided to do some digging—what exactly is vibe coding?

What is Vibe Coding?

After the video, I went straight to Google and searched What is vibe coding? Here’s what I found:

“Vibe coding” refers to a trend in software development where individuals use AI tools, like large language models (LLMs), to generate code by describing their ideas in plain language. This shifts the focus from manual coding to guiding and refining AI-generated code. Here’s your revised text with grammar fixes while keeping the flow and tone intact:

Why is Vibe Coding Taking Over?

We’ve seen how AI is changing the software development landscape, and Vibe Coding is just one example of how it’s being used. Heck, I’ve used AI to develop apps and solve a lot of errors that occurred in the codebase. I use AI to generate docs, create tests—but if all of this is happening, why is it trending everywhere? On X, YouTube, and across the entire software development community?

The main question is: why is this term suddenly being used so much, and why is it being called a bad thing? The interesting thing to me is that just a month ago, everyone was using AI, and then all of a sudden, everyone is against it. I think it’s because a few marketing people and tech CEOs are pushing the idea that AI coding is the future.

Deep Search

I decided to do some deep research, and I came across four interviews where CEOs from different AI companies are claiming that Vibe Coding is the future.

Here they are:

• Y Combinator
• Claude CEO
• OpenAI CEO
• Meta CEO

I’ve watched these videos, and in each of them, in some form or another, the CEOs are saying that Vibe Coding is the future. I have mixed feelings about this. This is happening because of marketing—they are pushing this idea so that developers spend money on their services.

Things like AI code editors, chatbots, AI agents—all these services are designed to hook new developers, making them so dependent on AI that they can’t leave it.

The Thoughts

Before I tell you how to do Vibe Coding, you should understand the difference between programming and coding. Sure, these things might seem the same, but under the hood, they’re totally different.

Coding is the process of writing instructions for a computer to execute. It’s like giving a robot a set of steps to build a house. Programming, on the other hand, is about designing how the house will be built—it’s the combination of writing instructions and the logic behind them.

Programming is the brain of the house—the blueprint that tells the robot how to build it from scratch.

The Difference

Here’s a table highlighting the key differences between coding and programming:

Aspect	Coding	Programming
Definition	Writing code that computers can understand.	The broader process of developing a complete software solution.
Scope	Focuses on translating logic into a specific syntax (writing code).	Involves planning, designing, debugging, testing, and maintenance.
Skills Needed	Basic knowledge of syntax and logic.	Requires problem-solving, algorithms, data structures, and system design.
Tools Used	Code editors (VS Code, Sublime Text).	IDEs, version control (Git), compilers, debuggers, and testing frameworks.
Complexity	Less complex, mainly syntax-focused.	More complex, involving architecture, debugging, and optimization.
Output	A working piece of code or script.	A fully functional software or application.
Example	Writing a simple Python script to print numbers.	Developing a full web application with frontend, backend, and databases.

In short, coding is a subset of programming—programming is the entire process, while coding is just one step of it.

The Problem

Not too long ago, I wrote an article called My Love and Hate Relationship with GitHub Copilot, where I talked about how this AI train has made me both a better and worse software engineer.

The core issue is that as new developers enter the industry, they might not learn how to do the basics. They don’t understand how to debug an app, how to make good decisions, how to identify problems. They might know what to do, but they’ll just throw in some prompts and hope for the best—getting stuck in a loop where AI makes decisions for them, and the cycle never ends.

They aren’t learning how to solve problems. They don’t understand how to think. They’re just copy-pasting code from the internet and hoping it works.

If you’d like to read my blog post, just click here. I also made a video about this and saw this problem coming a mile away.

The Solution

There is no perfect solution to this problem, and we can’t stop it from happening. I saw a person on X who used AI to create a game and made millions, but a few days later, the app was shut down because of a bad codebase—he didn’t know what to do.

See, without understanding core concepts, this will destroy your career and your skills.

But if you still want to use AI for your upcoming projects, I have a solution for you. When using AI to build products or write code, keep these points in mind:

1. Choose the Right Stack

• Pick a stack that is simple and one you understand well—something like React, Express, Tailwind, Redis, Deno, or even Node.js. The reason is that these LLMs are good at solving problems that have already been solved before.
  They are just autocomplete tools, not problem-solving tools.
• So if you go with something new, like a tech stack that was launched a month ago, it’s going to kill your vibes.
• Here’s the thing: AI is trained on Stack Overflow, GitHub, and Reddit databases. If you’re using a brand-new stack, it won’t have enough data to solve your problem.

2. Please Learn Git and GitHub

• If you’re going to use AI and embrace it fully, you must save your code before everything goes to hell. If you still don’t know Git, please learn it from my blog post.
• This will make things so much better because you can always go back to a previous version of your code if something goes wrong.

3. Think Like a Project Manager

• When using AI, you’re not just writing code—you’re building a product. You need to think like a project manager. Break your project down into smaller parts.
• Each section should have all the details AI needs to understand it better, but you should be the one making the decisions—not the AI.
• Be very specific and provide context because AI will forget what you’ve been working on over time.

There’s a new term for this called “Prompt Engineer.” I hate this term so much. I’ve seen it all over the internet, and it’s just a fancy way of saying, “I use AI to write code.”

If you want to learn more, check out this awesome video by Fireship on YouTube:

How Vibe Coding Works

Vibe coding is all about the back-and-forth between you and AI. You give instructions in plain English, and the AI translates them into working code. It’s like having an overpowered autocomplete that doesn’t just predict words but entire functions, components, or even full apps.

Here’s how the process usually plays out:

1. Describe what you want – Say you need a simple weather app. Instead of writing out all the HTML, CSS, and JavaScript yourself, you just tell the AI:

   “Create a web page that lets users enter a city and shows the current weather.”
   This is your prompt.

2. AI generates the code – Based on your description, the AI spits out a rough version of your app. It might generate the HTML structure, write the JavaScript for fetching weather data, and even add basic styles. It’s like having an assistant who just gets it (well, most of the time).

3. Review and refine – The first draft isn’t always perfect. Maybe the design looks bland, or there’s no error handling. So you tell the AI:

   “Make it more colorful and add error handling if the city isn’t found.”
   The AI tweaks the code accordingly.

4. Repeat until it’s right – You keep refining through quick iterations—describe the change, get updated code, test it—until everything works as expected.

Final Thoughts

AI is a powerful tool, but it’s just that—a tool. It can assist you, speed up your workflow, and help you solve common problems, but it cannot replace true problem-solving skills, creativity, and deep technical knowledge.

If you rely too much on AI without understanding the fundamentals, you’ll be stuck in an endless loop of prompting and hoping for the best. The best developers don’t just write code—they think critically, design solutions, and make informed decisions.

So, use AI wisely. Let it assist you, but don’t let it think for you. Stay in control, keep learning, and make sure you are the one driving your career—not the autocomplete.

Until then, peace out, nerds. 💕