For years, Balaji Vijayaraghavan, from Coimbatore in Tamil Nadu, helped founders launch all kinds of smartphone apps into the market. This gave him enough confidence to write a book on go-to-market (GTM) strategies, which is how you put your product out in a way that gives you the best shot at success.
But launching, as Vijayaraghavan found out, was different from building.
In early 2023, he cobbled together a team of 18 and a budget of ₹2 crore to build a GTM app and a market research automation tool. After two and a half years, neither was close to completion and only enough money was left to last eight months.
Resuscitation came in the form of vibe coding. Less than a year since then, both the products are live, built entirely on Emergent, a vibe coding platform.
Vibe coding, like several other technological revolutions, started as a thing of wonder and curiosity fuelling hobby projects. We heard stories about children building cute little things, such as nine-year-old Abdullah in Las Vegas vibe-coding his way to a chess app that makes the game less daunting for children like him.
This is possible because, in vibe coding, what you need to do is send prompts—voice or text—to artificial intelligence (AI) assistants to generate, refine, or debug software and apps. This changes the entire approach to software development by using language and speech to express an intention. AI transforms this intention into executable code by acting as the coding assistant, making suggestions in real time, and automate tedious processes. And, of course, everything is much faster.
Tasks that once required a developer, a roadmap, and a few weeks of wrestling with the first version of an app can now be done through a single prompt. For years, building the first version of an app took one to four months. Today, with vibe coding platforms, you can do it while your colleagues are having a coffee break.
To vibe code, you need not be a trained software engineer; what you need to be is good with language, speech, and prompts—and it does not matter if you are all of nine, like Abdullah of Las Vegas.
Now, though, vibe coding is expanding way beyond the hands of nine-year-olds. What we are seeing now is its largescale application in building software and apps and becoming critical tools of business while, at the same time, raising profound questions about the future of jobs and industries. The story goes that some of the best engineers in AI startups are not even engineers. So, what are engineers for?
Over the past year, every large and small information technology (IT) services company has adopted it, as have technology startups—each driven by the urge to build, test, and ship faster.
India’s $280 billion tech services industry now views vibe coding as a productivity lever. TCS has given its 600,000 employees access to top AI models. Infosys developers have already generated more than 7 million lines of code via GitHub Copilot. Cognizant says 20 percent of its code is now AI assisted.
All this is happening at breathtaking speed.
In June of last year, Hexaware hired people from colleges to train its management team in vibe coding. “Now all our management team can vibe code. Okay, not real code. But all of us can vibe code. It’s a little bit empowering. You suddenly find you can do things that you thought you couldn’t,” he says.
Hexaware launched a vibe coding offering in July last year. And the way Srikrishna describes the outcome is: “10x faster, 3x cheaper.”
Emergent, the vibe coding platform on which Vijayaraghavan built his products, was launched in June 2025. It claims to have nearly 6 million users, of which 150,000 are paying users across 90 countries, and to have crossed $100 million annual recurring revenue (ARR) in the eight months since its launch. ARR is the predictable revenue a company expects to earn in a year.
Rocket.new, another vibe coding platform, has raised $15 million. Composio raised $25 million, amid a surge in funding for vibe coding startups.
Globally, vibe coding platforms such as Lovable, Replit, and Cursor (Anysphere) have clocked multi-billion-dollar valuations and multiple rounds of funding over the last two years, with a corresponding spike in ARRs.
Emergent was born from the concept of “can we truly democratise software development—give every individual the power of a development team in the cloud?” Its own data says less than 1 percent of the world produces software currently. Vibe coding allows this to be extended to the other 99 percent.
“We believe every niche has a niche that needs software,” says Mukund. “In the last 30 years, most economic growth has come from software. We are limited by the number of people who can produce software.”
On launch day, he met a user who had built a simple swipe based app to help her and her spouse agree on a movie to watch together. She had no coding background. “You hear these ideas and think that these are such common problems, but no one ever thinks about building a solution for them. Even if they did, all this while it would have meant hiring a developer to make it,” says Mukund.
DhiWise, which started with code generation in 2021, pivoted to vibe coding in June 2025 with its new identity, Rocket.new. “In just eight weeks, from the first line of code to the first paying customer, we had gone live,” says co-founder and COO Deepak Dhanak.
Three weeks after launch, investor conversations began. Seven months later, it has crossed 1 million users in 180 countries.
The pivot, says Vishal Virani, co-founder and CEO, “was about recognising that the next generation user wants a vibe in everything.” He is quick to point out that Rocket.new is not just a vibe coding platform. Anyone can now generate software using Claude, GPT, or any vibe coding tool. The bigger challenge, he says, is knowing what to build.
Ninety-seven to 98 percent of vibe coded applications are never deployed, because, Virani says, “The question is not whether you can build something, but whether you should build it.”
“A year ago, for one of our clients, we created a software for $2 million, saving them $10 million for the next five years—so one year of return on investment,” says Srikrishna. “If we were to build it today, it would cost $0.5 million due to the acceleration in software development.” This would reduce reliance on SaaS (Software as a Service). “The line between services versus platform businesses is blurring,” he says.
The need to understand the code base is the key to vibe coding. Using LLMs to discover an existing code is not the solution. Using coding language like Python for discovery is a basic guardrail to ensure that vibe coding works for enterprises without hallucinating.
Instead of long documentation cycles and delayed feedback, Hexaware invites clients to sit for four hours and watch a working prototype take shape in real time. The client can sign off on the requirements the same day. This, Srikrishna says, is nothing short of transformative.
Rocket jointly built a fully integrated prototype for a client in about four weeks for a few thousand dollars—work the client estimated would take eight to 12 months and at least $10 million using the old way.
In the old model, business teams wrote requirements, product managers translated them, and specialised developers handled front-end, back-end, and deployment. “Everyone spoke a different language, so things kept bouncing back and forth,” says Abhijeet Kumar of TableSprint, a vibe coding platform.
Vibe coding is merging the roles of product manager, developer, designer, and tester into one. “A developer can think of an idea and ship it end to end. A designer can contribute directly to the codebase,” explains Mukund of Emergent. A large part of the work is likely to become automated, freeing people to operate at a higher level.
Expertise shifts from being defined by technical specialisation to domain understanding. “Earlier, the expert was the one who knew iOS or backend,” Kumar says. “Now the expert is the person who can explain the problem best.”
With vibe coding becoming increasingly mainstream, educational institutions may also need to rethink their curriculum. “Previously, we taught syntax first and architecture last. Now, we must teach architecture first. If AI can write the function, the student’s primary value lies in determining where that function belongs in a distributed system,” says Yash Sinha, assistant professor, department of computer science and information systems, at BITS-Pilani.
As vibe coding speeds up drafting, it raises the bar for evaluation, modification and risk assessment. V Krishna Nandivada, faculty, department of computer science and engineering, IIT-Madras, says: “It means that, in addition to building fundamentals, the courses must include code reviews, where students critique AI outputs, not just generate them. The goal is to produce engineers who not only depend on AI, but also supervise it.”
But having the tool does not make everyone a great builder. “Even with ChatGPT, you don’t become a good writer automatically,” says TableSprint’s Kumar. The core skill isn’t coding; it’s knowing what to solve and how to solve it.
Currently, for a lot of non-coders, most usage today is “prototyping, basic websites, ideation”, and on the 0 to 100 journey of building software, vibe coding solves “roughly 0 to 20”, explains Virani of Rocket.
Adoption in vibe coding improves dramatically when the prompter understands systems and constraints. Even now, a developer might be able to use vibe coding platforms a lot better than a non-coder. Mukund of Emergent says there is a need to close the education gap—both by simplifying the product and by training people through workshops.
Large datasets containing both good and bad code are used to train AI technologies. The biggest risks are security bugs, hidden logic errors, and AI repeating unsafe patterns learned from public code. According to industry surveys, a significant portion of AI-generated code has vulnerabilities. “This supports the idea that these technologies are accelerators rather than quality gatekeepers and that a human is needed to review the code before it is deployed,” adds Chhabra.
Therefore, the biggest differentiator among vibe coding platforms is how secure they are. “When we say we are built for enterprises, that means companies like Flipkart expect us to be secure by default,” says Kumar of TableSprint.
When Y Combinator President Garry Tan suggested that SaaS firms like Zoho could be “competed away” by vibe coding, Vembu asked why “email or spreadsheet or accounting apps” built through vibe coding did not already exist if the technology was so capable. He warned that without strong protections, vibe coding “just piles up tech debt faster and faster until the whole thing collapses”, and emphasised that enterprise grade software depended on guarantees that AI generated code could not yet provide reliably.
For Vembu, vibe coding may be a powerful accelerator, but not a replacement for depth, integration and rigour. “The idea is to use vibe coding for basic queries, build a context window, then go for a different model to refine the output. If your ask is as simple as adding a button on a page, it is easy to deploy. If it has to be structured and standardised as per the rest of the site, then you need to work with it,” says a developer working on an academia project for a research lab who is not authorised to speak publicly.
As things stand, vibe coding will not convert everyone into a software builder overnight. But everyone can have a shot at it. And that includes nine-year olds.
First Published: Mar 19, 2026, 16:15
Subscribe Now