What is Vibe coding?

Home

What is vibe coding? Besides the new polarization of society? You must have registered the name somewhere on LinkedIn, YouTube or some forum. Plus it's already on Wikipedia, so we have to join in.

Vibe coding is a way of creating applications through AI where prompting is used instead of manually writing code. Vibe coding relies on tools such as GitHub Copilot, Cursor, Replit Agent or ChatGPT to interpret prompts and create code - from simple prototypes to web apps.

This can speed up development, but only under the supervision of an experienced programmer. However, it could free their hands for more important activities - such as testing or ensuring the security of the application.

How did it start?

The term "vibe coding" comes from , a prominent AI researcher and former co-founder of OpenAI from February 2025.

The tweet begins like this: "There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists."

Translated, "There's a new kind of coding I call "vibe coding" where you fully surrender to feelings (vibes), embrace exponentials, and forget that code even exists."

What can vibe coding be good for?

Speed and productivity

For senior developers, this can save time. Thanks to AI, they can now focus on solving more complex problems instead of boilerplate (repetitive code that is mostly the same across projects) code or syntax. AI takes care of routine tasks like form creation or minor refactoring, allowing developers to focus on things that require complex logic.

Prototyping

For non-technical or design positions, there is the opportunity to quickly prototype an application or parts of an application. Figma, for example, also offers prototyping, but an interactive site can be far more interesting and useful (for UX improvements, for example).

Explorations

The conversational nature of vibe coding encourages creativity and allows developers to explore ideas they might not have considered before. There are even voice-controlled tools such as that enable a hands-free coding experience.

Where can there be a problem?

Code quality

AI-generated code may contain bugs, be inefficient, or may not be completely secure. LLMs may produce code that works superficially but hides problems that may not be immediately obvious. A well-known companythat AI-generated code is often vulnerable to problems such as SQL injection (a vulnerability where a user can send SQL to a server and change/delete data there) and should always be checked.