A liberal-arts-trained product planner launched a running-habit app called "Routinist" on the App Store without sitting through even an hour of a coding course. From the first day he started working to the day it launched, it took five weeks. The tool he used was an AI coding-assistant environment, and what he did was describe what he wanted in plain language and check the results. The code was generated automatically in between.

He summed up the experience as "a strange and delightful world." What stands out about that reaction is that it was framed in the language of astonishment rather than accomplishment. It also means he hadn't expected that building an app himself would be possible at all. Before that astonishment scatters, there's reason to look more closely at what this case actually holds.

Why this approach has started to work now

The term "vibe coding" became widely known through a post that AI researcher Andrej Karpathy put on social media in early 2025. The idea was to work by accepting the AI's suggestions without trying to fully understand the code. The method was simple. You describe what you want, the AI generates code, and if it doesn't work you paste the error message back into the AI, check the result, and shape your next request. The claim was that even without any development knowledge, repeating this loop alone would produce working software.

At the time, the developer community was largely skeptical. Is it acceptable to deploy code you don't understand? How do you fix things when an error occurs? Who is responsible when a security vulnerability appears? Those questions are still valid. The debate hasn't closed.

But while the debate played out, the quality of the tools changed. Over the course of 2024 and 2025, AI code-generation tools grew noticeably better at producing code that actually works. Environments like Cursor, Lovable, and Bolt.new evolved by integrating a code editor with an AI chat window. The flow of passing an error message straight to the AI and getting corrected code back stabilized. This iterative process began to connect all the way through to a finished product, and Choi Cheol-yong's "Routinist" is one case where that possibility became real.

What to watch out for inside this story

It would be a mistake to read this case and jump straight to the conclusion that "I could build an app in five weeks too."

Choi Cheol-yong was a product planner. He was already someone who could put into words what an app should do, what flow users should move through, which features were necessary and which were excessive. In vibe coding, the quality of the "description" you hand the AI determines the quality of the output. A specific, consistent description yields consistent code; a vague request yields a vague result. Reproducing the same outcome with AI tools alone, without planning skill, is hard. Behind his finishing a product in five weeks lies the fact that he could describe what he wanted in concrete terms.

The scope of "Routinist" deserves a look too. It's a relatively simple app built around a single purpose: tracking a running habit. The story is different for a service that processes financial transactions or handles large volumes of user data. In situations where code you don't understand could become a security vulnerability or lead to the loss of user data, vibe coding is not a sufficient method. In that context, you need the code reviewed by a developer or a separate security audit.

There was a similar discussion when 3D printing spread. Anyone could now print physical objects themselves, but in safety-critical areas like medical devices or load-bearing structures, professional design processes were still required. It's the same way not every publication earns the same trust simply because printing technology became widespread. Being able to make something easily and being allowed to make it for any purpose are two different matters. Vibe coding is the same: the barrier to entry being lowered and that barrier disappearing are not the same thing.

What a solo operator can genuinely take from this trend

So where is the context in which this method actually works in a meaningful way?

Internal tools, or projects built for yourself, are the most realistic starting point. When something isn't released publicly, or its user base is just you and a handful of others, whether it works matters more than how polished it is. Automation scripts that cut down on repetitive manual work, a simple dashboard that pulls data from several platforms into one place, a tracking tool tuned to your own work rhythm — these all fall into this category. Outsourcing such tools costs anywhere from a few hundred thousand to several million won, and every revision request adds communication overhead. Building them yourself changes that cost structure.

It's also valid for an early prototype meant to test market reaction. At the stage of quickly building a version made up of only the minimum features and putting it in front of real users, speed matters far more than polish. At that stage, vibe coding is a way to move faster than waiting on a developer.

For those who want to actually get started, it helps to write out a scenario before writing a feature list. If you lay out, in narrative form, what the user sees when they open the app, where they tap, and what they'll want, the density of the description you hand the AI changes. The feature list can wait until after the thing is finished.

An attitude of not fearing errors matters too. In the vibe coding process, an error isn't an accident to be avoided but part of an iteration you solve together with the AI. If you copy the error message verbatim and pass it along with the context — "I got this error, how should I fix it?" — a direction usually emerges. This process is itself the workflow of vibe coding.

Some people point to this trend and say "now everyone's a developer." I don't think that phrase captures the change accurately. It's less that coding has become universal, and more that one context has emerged in which a planner no longer has to wait for someone who writes code. A possibility has opened for someone who can describe what they want to make to put out a product themselves, and "Routinist," which Choi Cheol-yong placed on the App Store in five weeks, is confirmation that the possibility genuinely works.