Reference · 01 of 06
What is vibe coding?
Vibe coding is the practice of building software by describing intended outcomes in natural language to an AI, which then generates and iterates the necessary code. The developer directs this process by defining requirements and evaluating results rather than writing most lines of code by hand.
Where did the term come from?
The term was popularised in February 2025 by Andrej Karpathy, an AI researcher and former director of AI at Tesla. In a widely shared post, he described coding by giving in to the vibes and accepting AI-generated code with light review, rather than checking every single line manually.
What vibe coding is, and is not
Vibe coding involves generating larger sections of code from a description of the desired outcome, rather than just accepting autocomplete suggestions one line at a time. It is not an invitation to skip review entirely; responsible use still requires testing and checking the result. The defining trait is a shift toward reduced line-by-line authorship, not the total removal of oversight.
Who is vibe coding for?
It is primarily for people prototyping quickly with little or no programming background, such as founders, hobbyists and non-technical product managers. It is also used by experienced developers to speed up scaffolding, boilerplate and initial prototypes. It is less suited to production systems that require strict security or correctness guarantees without careful, expert review.
How is it different from traditional coding?
In traditional coding, a developer writes and edits code directly and maintains an understanding of every line, with AI, if used at all, limited to small, discrete suggestions. In vibe coding, the developer describes the desired behaviour and the AI handles the implementation. The developer's primary skill shifts from typing syntax to specifying intent clearly and evaluating the resulting output, the mechanics of that loop are covered in how vibe coding works.
How is it different from no-code and low-code?
No-code and low-code tools use prebuilt visual components and form-based configuration, which restricts the builder to a specific platform's building blocks. Vibe coding produces actual source code in a real programming language from natural-language prompts. The capability of the output depends on the underlying AI model rather than a fixed library of visual components, so it is not bound to any one platform's toolkit.
Common questions
- Does vibe coding work for every project?
- It is most effective for prototyping and rapid development, but it carries risks for complex production systems where strict correctness is required.
- Is vibe coding the same as using an AI autocomplete tool?
- No. Vibe coding involves generating larger stretches of code from a description of the outcome, whereas autocomplete tools generally offer single-line suggestions while you continue to write the code yourself.
- Do I need to be a developer to use vibe coding?
- You can use it to build prototypes without a programming background, though understanding the basics helps when reviewing the generated code.
- Where can I learn about the specific tools used for this?
- A survey of platforms such as Lovable, v0, Bolt, Replit, Cursor and Claude Code is covered in the tool landscape page on this site.