What is ‘vibe coding’ and how you too can jump on the bandwagon

Up until a couple of years ago, software development felt like a world reserved for that bunch of nerds who understood the difference between Java and JavaScript, or who could navigate dense documentation without breaking a sweat.

Then, as AI discussions picked up pace, a phrase entered the scene: vibe coding. At first, it sounded like a joke, a playful way to describe letting artificial intelligence "do the coding while you vibe".

But now, it is a mark of a cultural shift in how people build software, who gets to build it, and what creativity looks like when machines help carry the load.

What is vibe coding?

The rise of vibe coding begins with a simple idea.

What if you could describe the app you want in plain language, and an AI system could write the code for you? No syntax. No semicolons and brackets. Just natural conversation.

The term was coined earlier this year by Andrej Karpathy, the former Tesla AI lead and OpenAI co-founder, who framed the idea as building software through "vibes, not variables". It was both humorous and provocative — and the internet latched onto it quickly.

Even Collins Dictionary recognised the cultural moment and named "vibe coding" its 2025 Word of the Year.

Vibe coding took off because it feels intuitive. People already speak to AI assistants daily, asking them to summarise articles, suggest recipes or draft presentations. Extending that approach to software development was a natural next step.

A developer might ask for an app with a login page and a database. A student might prompt an AI to create a website that generates holiday gift ideas. What used to require hours of labour now takes only a few minutes.

Rise of 'vibe coding startups'

This shift has also coincided with the rapid rise of AI-native startups built on this philosophy. The story of Giggles, an app created by a team of teenagers, is often cited as a turning point.

The founders built the entire platform, from creative features to interactive digital collectables, by orchestrating AI tools rather than hiring a conventional engineering team.

Another example, Base44, grew to hundreds of thousands of users and sold for 80 million dollars within just six months. And the fun part is: neither of these companies started with teams of experts.

Vibe coding took off because it feels intuitive. People already speak to AI assistants daily, asking them to summarise articles, suggest recipes or draft presentations. Extending that approach to software development was a natural next step.

Big Tech has taken notes as well. Google, Salesforce, OpenAI and others are racing to launch their own agentic coding systems. Platforms like Lovable, Cursor and Replit have grown at extraordinary speeds.

The elephant in the room

This enthusiasm surrounding vibe coding comes with some caveats as well. The relaxed optics of vibe coding can be misleading. These tools can sometimes generate faulty code or misinterpret instructions. Developers who understand the fundamentals still have an advantage in steering the process.

A beginner can build a prototype in an afternoon, but scaling it, securing it and maintaining it is another story. As GitHub's chief executive observed recently, no founder can grow a company without engineers who understand what lies beneath the surface.

This tension defines the current phase of vibe coding. On one hand, it democratises creativity, giving a wider range of people the chance to build. On the other hand, it raises questions about reliability, cost and long-term sustainability.

Many of these tools rely on expensive foundational models supplied by companies like OpenAI or Anthropic. Some startups report negative gross margins because the cost of generating code exceeds subscription revenue. And while rapid prototyping is easy, many users encounter limitations when they attempt to build deeper, more complex products.

Yet, vibe coding remains exciting because it captures a shift in mindset.

Coding is no longer only about knowing the rules of a language. It is increasingly about knowing what you want, how to describe it and how to collaborate with an intelligent system. The process feels more like sculpting than engineering: you begin with a rough form, refine it through conversation and shape it toward a final design.

So how do you jump on the bandwagon?

Start small. For instance, imagine you want to build your own website. You could open an AI coding tool — ChatGPT, Google AI Studio, or Claude — and type something simple, like, "Create a personal website with a homepage, an about section, and a contact form."

The AI will generate a draft for you. You can then tweak colours, fonts, layout or add new sections by simply describing the changes in plain language. If the AI makes a mistake, ask it to fix it.

Experiment with different prompts, test your site, and gradually refine it. You do not need to know HTML or CSS in depth to get started — vibe coding lets curiosity guide the process while the AI handles the heavy lifting.