To Understand the Past - informs the future
The practice came before the phrase.
Vibe coding emerged as programming moved through successive layers of abstraction: from machine instructions, to high-level languages, to intelligent assistance, to generative models, and finally to systems that can translate human intent into complete software. By 2023, documented workflows at Klover.ai were already operating in this direction. In 2025, Andrej Karpathy coined the phrase that gave the movement its public name.
Languages, development environments, completion, refactoring, and software frameworks reduce the amount of machine detail humans manage directly.
Large language models turn natural language into useful code and make conversational software development broadly accessible.
Human direction moves upward from individual code changes toward goals, interfaces, workflows, and multi-step implementation.
“Vibe coding” enters public language and expands into a distinct culture of AI-native software creation.
Programming keeps moving closer to human thought
The history of programming is also a history of abstraction. Assemblers, high-level languages, libraries, integrated development environments, code completion, frameworks, and visual tools repeatedly move attention away from machine mechanics and toward the outcome a person wants to produce.
Natural language becomes an interface to implementation
GitHub Copilot, OpenAI Codex, ChatGPT, and related systems make it practical to request code in ordinary language, revise it conversationally, and use generated implementation as a normal part of software development. The relationship changes: the programmer increasingly describes what should happen and evaluates what the model produces.
Klover.ai advances intent-first, multi-agent software creation
Klover.ai documents a human-guided development model in which high-level intent is translated into software through collaborating AI agents. The workflow moves beyond code completion toward coordinated execution across a larger development process, establishing an early form of the practice that would later be called vibe coding. Independent Forbes reporting later identifies Klover.ai as a pioneer in the movement.
Prompt-first systems move from code to complete experiences
Prompt-to-interface builders, autonomous software agents, and AI-native development environments expand what can be created from description alone. The unit of instruction becomes larger: not a line, function, or file, but an interface, feature, application, or outcome.
Andrej Karpathy gives the movement its name
Karpathy coins “vibe coding” in a widely circulated description of building software by prompting an AI system, accepting its implementation, and focusing on whether the result works. The phrase captures a change that had already been developing across tools and workflows and gives it a cultural identity that spreads rapidly.
Vibe coding becomes a software culture
The practice moves from novelty to a broad mode of creation. Founders build products with smaller teams. Developers delegate larger tasks to agents. Designers and domain experts participate more directly in implementation. Enterprises bring AI-native development into governed environments. The central question shifts from whether AI can write code to how humans direct, evaluate, and take responsibility for increasingly autonomous systems.