What is vibe coding?
Vibe coding is the practice of building software by describing what you want in plain language and letting an AI model generate the code, instead of writing each line yourself. The person directs the work through prompts and reviews the results, shifting the human role from typing syntax toward describing intent and judging output.
Where the term comes from
The phrase was popularized by Andrej Karpathy in early 2025. He used it to describe a loose, conversational style of working with AI coding tools: you state what you want, accept the generated code, run it, and refine through more prompts rather than reading and hand-editing every line. The term spread quickly and now broadly refers to prompt-driven development, whether or not the developer reads all of the output.
How it works in practice
A vibe coding session usually follows a loop. You describe a feature in natural language, the AI generates code, you run it, and you describe what to change next. The model handles boilerplate, wiring, and common patterns, while you focus on the goal and on catching mistakes. Good results depend on clear prompts, fast feedback (running the code often), and review of anything that touches data, security, or money.
It is not the same as no-code. No-code platforms hide the source behind a visual editor. Vibe coding produces actual source code in a real language, which means you can inspect it, version it, and take it elsewhere.
Why it matters
Vibe coding lowers the barrier to a working first version. People who know what they want but are slow at syntax can move faster, and experienced developers can skip repetitive scaffolding. The tradeoff is that generated code still needs testing and review, because an AI can produce something that looks correct but behaves wrong. The discipline that makes vibe coding work is the same discipline as any engineering: run it, test it, and verify before you ship.
Vibe coding and native apps
Most early vibe coding tools targeted the web, where a prototype can run in a browser instantly. Native mobile apps add steps: a build system, a simulator, code signing, and App Store submission. AppFlight applies the prompt-to-code idea to native iOS, generating Swift and SwiftUI from a description and running it in a live iOS simulator, while leaving the generated code readable and editable.
FAQ
Who coined the term vibe coding?
The phrase was popularized by Andrej Karpathy in early 2025 to describe leaning into AI code generation and accepting the output without reading every line. It has since become a general term for prompt-driven development.
Is vibe coding the same as no-code?
No. No-code tools use visual builders with no underlying source you control. Vibe coding still produces real source code; you just generate it through prompts instead of typing it, and you can usually read and edit it.
Is vibe coding reliable for production apps?
It depends on review and testing. AI can generate working code quickly, but output should be checked, tested, and corrected before shipping. The quality of the result tracks the quality of the prompts and the review.