Y Combinator CEO Garry Tan is convinced that vibe coding will transform how startups are built, as it makes companies far more efficient and cost-effective. Let’s unpack it.

What is This Vibe All About?

Imagine: you have an idea for an app. Before, there was only one path — hitting the books on Python, Java, or something else, spending months learning all about functions, classes, and libraries. Or hiring a developer, which is expensive.  Now you can fire up ChatGPT, Claude, or Overchat AI and ask: “Make me an app that translates any language to another language in real-time.” The AI thinks for a bit and spits out ready-to-use code.

The term vibe coding was introduced by OpenAI co-founder Andrej Karpathy in February. It refers to the active use of AI tools to build software — and right now, it’s generating a lot of buzz in the tech world and on social media.

Karpathy described it as a process where you fully rely on “the vibe,” on intuition, and forget about the code itself. You just set the direction, catch the wave — and the machine writes. The idea instantly resonated. Suddenly, millions of people with zero coding background could create things. It’s like being handed a magic wand for building digital stuff.

Is Vibe Coding Here to Stay?

Y Combinator CEO Garry Tan is convinced that vibe coding will radically transform the startup ecosystem, as it makes companies far more efficient and cost-effective.

“Startups are reaching $1 million to $10 million in annual revenue with fewer than 10 people. That’s never happened before at such early funding stages,” he noted in an interview with CNBC.

“You just talk to large language models, and they write entire applications for you. And if not — if there’s a bug, or you want to make changes, or change the way it looks — you don’t have to dive into the code and write it yourself,” Tan explained. This speeds up and simplifies development, he continued. Plus, the approach makes it more justifiable to spend resources on building niche software for highly specialized or small markets.

How Big is the Productivity Boost?

What used to take a team of 50 or 100 people can now be handled by 10 skilled vibe coders. But Tan emphasized that to be this productive, they really need to be highly proficient with the latest AI tools for developers.

And here’s where it gets interesting. The biggest strength of vibe coding is accessibility. It breaks down the barrier to entering the world of software development. How many brilliant ideas have been collecting dust just because their creators didn’t know how to code? Now they’ve got a shot. In a couple of evenings, through trial and error (using “prompts,” as they’re called), you can end up with a working prototype. For a non-programmer — this is mind-blowing. Something they could never have built themselves.

Are there Downsides to Vibe Coding?

A lot of time and effort currently goes into debugging the generated code and figuring out what it’s actually supposed to do. For now, humans still have to do the dirty work — LLMs aren’t quite good at that yet.

Earlier, the Y Combinator CEO mentioned that in the accelerator’s latest batch, 81% of startups are focused on AI. And for a quarter of them, nearly all of their code is AI-generated.

First off, the code that AI generates is often far from perfect. It might be inefficient, buggy, or contain security flaws. And here’s the kicker: the person who “vibed” that code probably has no clue how to fix it. All they see is the result — it’s either broken or janky. Now what? Go back to the AI and say, “Okay, now fix it”? That creates a feedback loop where you’re completely dependent on the same tool that caused the problem. Reliability is still shaky, to be honest.

Second, it raises the question of what programming even means. Writing code with a full understanding of each step, each line, the logic — that’s one thing. But getting a finished result without diving into the details? Is that really coding? Or is it more like advanced prompting or managing?

105 Birthday Wishes for Sister बहिणीला वाढदिवसाच्या शुभेच्छा in Marathi

So, What’s the Verdict?

Vibe coding is definitely an exciting development. It’s a vivid example of how AI is changing the way we work. It opens creative doors for people who used to be left out and could potentially speed up development. But for now, it’s more of a thrilling experiment than a full-blown replacement for traditional programming. There’s enthusiasm, there are early results — but also real limitations, quality concerns, and reliability issues.

So let’s keep watching. Maybe in a couple of years, these tools will be smarter, more stable, and vibe code will go from buzzword to everyday practice. Or maybe it’ll stay a niche solution for certain projects.

Read more related blogs on Marathi Quotes Hub. Also Join us WhatsApp.