Practical AI
Is Vibe Coding the New Coding?
Practical application of AI still requires code. All the frontier LLMs will sometimes generate Python or other code to achieve an objective prompted by a user. Anthropic is so convinced of this that it introduced Claude Code, which has been called an ideal platform for "vibe coding."
What is Vibe Coding?
The term "vibe coding" was coined by Andrej Karpathy, an OpenAI co-founder and former Tesla technology executive, and it gained rapid cultural traction, with Collins Dictionary naming it the 2025 Word of the Year. Web searches for "vibe coding" reportedly jumped 6,700% in spring 2025, reflecting its explosive mainstream adoption shortly after Karpathy's introduction.
Just to make sure we're on the same page, vibe coding refers to users describing desired application functionality in natural language and letting AI generate the code, rather than manually writing it line by line. It has quickly replaced low-code/no-code (LCNC) tiled interfaces as the preferred way for regular business users to build simple applications and workflows for their own work.
Claude Code, Anthropic's tool for non-coders, was recently used by the company internally to produce Claude Cowork, an application that provides a simple interface for users to request that the Claude LLM perform specific work functions. Reportedly, the application went from conception to delivery within 10 days, all of it vibe-coded.
Although vibe coding has enabled business users to create simple applications, complex full-stack applications still require the expertise of professional developers, particularly expertise beyond coding itself.
Professional Developer Adoption
Professional developers embraced AI-vibe coding tools almost immediately, and their use continues to grow exponentially. According to 2025 data, 84% of programmers have tried at least one AI code generator tool, with 82% using AI coding tools daily or weekly as part of their normal workflow. Approximately 41% of all code written in 2025 was AI-generated, and 61% of developers report that AI touches at least one-fourth of their codebase.
Full-stack developers lead adoption at 32.1%, followed by frontend developers at 22.1%, and backend developers at 8.9%. Nearly 76% of professional developers either currently use AI coding tools (62%) or plan to adopt them soon (14%). Among top engineering teams, usage rates reached 85-90% daily by the end of 2025, meaning AI assistants had become part of the software delivery operating system rather than a side experiment.
By December 2025, there were reports suggesting that "100% of contributions" from Claude Code creator Boris Cherny were now written by Claude itself, though this appears to refer to ongoing contributions rather than the entire codebase. It is also reported that Claude Code, Anthropic's command-line AI coding tool, has 80% of its own code written by Claude Code itself, according to Cherny. This figure was revealed during an interview on the Latent Space podcast in 2025.
Developer Perspectives on Effectiveness
Professional developers report significant productivity gains but emphasize the importance of engineering, architecture, and structured development expertise. One developer noted that using the AI development platform Cursor has made them "five times more efficient" while maintaining quality, though they cautioned that "true vibe coding, where you simply ask the model to create without adjusting the underlying architecture or code, is not yet reliable enough for production use." Another developer reported that "Ideas that sat in our backlog for years have come to life in just a week, thanks to AI."
However, experienced developers consistently warn about risks. Chris Reynolds, a software engineer at Pantheon, explained that "if you're just an average power user without any development experience, you could absolutely get AI to build you a cool thing that looks and acts the way you want, but if you don't know how any of it works, when it has issues or bugs or whatever, you can find yourself deeper and deeper in the rabbit hole."
Critical Skills That Differentiate Professional Developers
Technical Foundation and Code Comprehension
Professional developers possess deep knowledge of programming language fundamentals, data structures, algorithms, and system design principles that remain essential even in the vibe coding era. These foundational skills are rated as "high importance" for evaluating and refining AI outputs, ensuring proper architecture, and identifying issues. While AI can generate code quickly, developers need this expertise to "comprehend and refine AI-generated code so that it adheres to stringent criteria".
A developer on Reddit explained the practical difference: "I firmly believe a coder who knows a language can build something 100x better than someone who doesn't and uses AI" because understanding programming fundamentals enables them to recognize when AI relies on outdated APIs or suggests suboptimal approaches.
Code Quality, Security, and Debugging Expertise
Professional developers naturally excel at evaluating AI-generated code for accuracy, security vulnerabilities, and functional correctness before deployment. They possess systematic debugging abilities that normal users have never acquired. As one developer noted, "the non-coder has no idea how the code works. If there is a bug (which 100% will happen eventually), they will have to understand their own code in order to find and fix it, which takes the same amount of time as coding it in the first place".
Security knowledge is particularly crucial, as developers can identify common vulnerabilities such as SQL injection, path traversal, and hardcoded secrets that AI might inadvertently introduce. Willem Delbare from Aikido security warned that "vibe coding makes software development more accessible, but it also creates a perfect storm of security risks that even experienced developers aren't equipped to handle. SQL injections, path traversal, hardcoded secrets. Many developers consider security a blind spot, so imagine the risks when going from a simple landing page build to a full-fledged product handling user data."
Architectural and Systems Thinking
Developers bring established practices for the overall project design workflow, module and component integrations, and database structure planning. This systems thinking enables them to ensure AI-generated code fits coherently within larger architectures rather than creating isolated components that don't work together.
An experienced developer explained their advantage: "I think the biggest advantage I have as an experienced coder over someone who has no experience coding is having an established practice of how I work through specific common challenges" including project design workflow and knowledge of fixes used in the past that can be described to AI for implementation.
Prompt Engineering and AI Management
Professional developers have superior prompt engineering skills that optimize AI outputs through precisely crafted prompts with clear requirements, context, and constraints. They understand how to iterate on prompts, manage context windows, and route tasks to appropriate AI models based on specialization. Developers also excel at managing AI outputs, knowing when to accept, refine, or reject AI suggestions.
Domain and Operational Knowledge
Beyond pure coding skills, developers bring an understanding of deployment processes, version control systems, testing methodologies, and production environment requirements. They know how to track code modifications, maintain documentation, and implement security best practices throughout the development lifecycle. Professional developers also have knowledge of established fixes and workarounds accumulated through years of experience that they can articulate to AI for implementation.
Implications for Normal Users
Democratization vs. Quality Tradeoffs
Vibe coding significantly lowers the barrier to entry for non-programmers, enabling people without formal coding experience to build working software or prototypes. This democratization allows anyone to create apps in plain English, potentially unleashing innovation from unexpected sources.
However, this accessibility comes with serious caveats. The consensus among professional developers is that vibe coding "is not for newbies" when it comes to production systems. Security experts warn that "two engineers can now churn out the same amount of insecure, unmaintainable code as 50 engineers" because AI supercharges both development speed and risk.
The Cost Curve Problem
For normal users, the initial experience with vibe coding feels transformative. Scaffolds appear quickly, features materialize, and applications light up. However, the total cost of ownership can be a problem. "AI only" approaches start with low costs, but the curve bends sharply as projects increase in complexity, facing integration delays, security patching, and unavoidable refactors.
The most successful pattern is "Human + AI," where experienced oversight combines with AI generation speed. This means normal users can prototype and experiment effectively, but transitioning to production-grade software still requires professional engineering judgment for architecture, security, testing, and operational reliability.
Is Vibe Coding the New Coding?
Vibe coding represents a significant workflow shift rather than a complete replacement of traditional coding. The role of developers is "shifting upward, not disappearing" with engineers increasingly functioning as system designers, reviewers, and operators of AI-assisted pipelines rather than pure code writers. For normal users, vibe coding enables rapid prototyping and experimentation, but commercial software still demands engineering expertise for requirements translation, architectural coherence, security posture, and long-term maintainability.
The bottom line is that "vibe coding changes the workflow, not the need for engineers". While AI makes code production cheaper and faster, organizations still pay for outcomes such as reliable workflows, predictable releases, and controlled risk, which continue to require professional oversight.
About the Author
Technologist, creator of compelling content, and senior "resultant" Howard M. Cohen has been in the information technology industry for more than four decades. He has held senior executive positions in many of the top channel partner organizations and he currently writes for and about IT and the IT channel.