Vibe Coding

The term "vibe coding" was popularized by Andrej Karpathy in early 2025. He described a workflow where you "fully give in to the vibes, embrace exponentials, and forget that the code even exists." The idea is that AI tools have become capable enough that you can describe what you want conversationally and get working software without manually writing most of the code.

In a vibe coding session, you might say: "Make me a dashboard that shows user signups over time with a nice chart." The AI generates the full implementation. You look at the result, say "make the chart blue and add a date picker," and the AI updates the code. You iterate on the output rather than the code. The focus is on outcomes, not implementation details.

Vibe coding works well for prototypes, side projects, and MVPs. For production systems, you still need to review the generated code for security, performance, and maintainability. Shipping code you do not understand creates technical debt.

Vibe coding is most effective when the developer has enough experience to evaluate the output even if they did not write it line by line. Knowing what good code looks like matters—even when you are not the one writing it. The skill shifts from implementation to evaluation and direction.