Imagine giving voice prompts to your computer in plain English – and getting an app or software built in just a few minutes. This scenario is fast becoming a reality thanks to “vibe coding”, the latest buzzword in tech circles that has caught the imagination of software developers in the Silicon Valley in San Francisco Bay Area of California, as well as in Bengaluru. First coined by OpenAI’s cofounder Andrej Karpathy in February this year, vibe coding is a new AI-driven approach to software making that promises to accelerate projects and lower the barriers to creating software. It allows the coders focus on the creative aspects of development rather than getting bogged down in technical details. There are potential drawbacks too, but more on it later.
“Think of it like jamming with a friend: you can start with your idea, make changes, and iterate faster all while seeing your application come to life before your eyes,” Kyle Daigle, global COO of GitHub told FE. “Vibe coding is the latest way for both developers and citizen coders to build apps with AI taking their natural language prompts and going from idea to running app without the need to manually edit the code itself. India’s developers are already embracing this future,” he said.
This is not just about writing code faster—it’s about reimagining the developer experience, according to Srividya Kannan, CEO, Avaali Solutions. Tools like GitHub Copilot, Replit, and others are already demonstrating how AI can shorten the feedback loop, turning ideas into prototypes with unprecedented speed. At its core, vibe coding refers to gauging a workplace’s emotional tone or “vibe” using technology and data analytics, offering actionable insights to bridge gaps between organisational goals and human emotions. “While it may sound like a buzzword at first glance, the underlying idea is compelling: AI translating human intent into working code through ambient inputs, natural language, and contextual understanding,” she opined.
Several factors have converged to make vibe coding possible. First, AI’s ability to write code has advanced dramatically since late 2022, when tools like ChatGPT burst onto the scene. AI systems have become powerful enough to quickly take up complex requests, write complex code and even debug errors. At the core are advanced large language models (LLMs) like Anthropic’s Claude 3.7 and Google’s Gemini 2.5 Pro, which are among the top contenders for AI coding assistance.
“Until a few months ago, AI could only perform basic tasks like auto-completion. But with new advancements in AI models, such as Claude 3.7 and Google Gemini 2.5 Pro, developers can now provide screenshots or English prompts, and the AI can generate code modules accordingly,” said Paramdeep Singh, co-founder of Shorthills AI.
This is a major shift. What was earlier limited to minor suggestions is now full-scale assistance, where the developer acts like a pilot giving instructions, and the AI handles the execution. These models can write code, generate documentation, create test cases, and improve overall developer efficiency. Tasks that used to take days or weeks can now be completed in hours. “This changes how software development is approached and executed,” he added.
An obvious ramification is that engineers will now need to reskill and upskill. Previously, many could rely on writing boilerplate code and still remain relevant in the market. But now, LLMs can generate such code—thousands of lines in minutes. To stay competitive, developers must shift to solving more complex problems that require layered thinking, which AI cannot yet replicate. This includes understanding business needs and translating them into technical solutions.
Custom algorithms and non-standard workflows will still require human input. Singh stressed that engineers who haven’t moved beyond routine coding will need to evolve and focus on adding real business value.
Of course, fundamental coding skills will always be essential. At its core, coding is problem solving—that hasn’t changed. Even as AI becomes more capable, it’s still the developer who sets the direction, provides the context, and makes the final decisions. According to Daigle, developers need to guide the AI, review its output, and understand what the code is doing. AI can accelerate the process, but it’s the human insight, logic, and creativity that unlocks its full potential.
However, Kannan cautioned that it’s important to separate vibe from velocity. While these tools are powerful, they’re not infallible. Code generated without sufficient guardrails raises serious concerns around privacy, security, scalability, data ownership and long-term maintainability. There’s a real risk of creating beautifully written but functionally fragile systems if we don’t invest in rigorous testing, reviews, and governance.
Manish Shekhawat, director – experience engineering, Publicis Sapient, said that Indian developers are embracing vibe coding efficiently, recognising its potential to streamline workflows and enhance productivity. Yet, speed must be matched with responsibility. Security, maintainability, and code quality remain non-negotiable, he summarised.
