Building a Website Without Ever Learning to Code
A simple project ultimately became a lesson in the difference between making something look good and actually taking it all the way to something people can use.
A simple project ultimately became a lesson in the difference between making something look good and actually taking it all the way to something people can use.
Can someone without a programming background actually build a website from scratch and take it all the way to production with AI as a partner along the way?
Earlier this year, a lecturer from the Faculty of Economics and Business at Unpas approached me to build a website for an academic program as part of its accreditation requirements. The brief was simple: a landing page containing key information, the program's profile, a brief history, faculty members, and contact information.
My background is not in programming. I studied Visual Communication Design, and most of my daily work revolves around layout and graphic design rather than code. But that seemingly simple request became an opportunity to step outside what I was used to: building the website myself, with the help of AI.
The landing page was completed within a few days. I chose React and Tailwind CSS and built it with AI assistance, then hosted it on Hostinger using the most basic plan. A week later, the website was live.
What I did not expect was that this was only the beginning.
As the accreditation process progressed, the requirements grew. The client asked for a news section, individual faculty profile pages, and several other adjustments. For someone whose primary skill is design, these were not small requests.
What had started as a simple static page now needed a data structure, dynamic pages, and a way to manage content as it continued to grow.
This was where I started learning things that were genuinely new to me: how databases work, how the backend connects data to the interface, and why a website's architecture can vary significantly depending on its requirements.
One technical limitation ended up shaping the direction of the entire project. The hosting plan I was using did not support Node.js, which meant the modern framework options I had initially considered were no longer viable.
The remaining option was PHP and MySQL—two technologies I had heard of before but had never actually learned.
Migrating from a React landing page into a complete PHP and MySQL system became the most challenging part of the project.
It was not just about learning the syntax of a new programming language. I had to understand how the system worked as a whole: how data was stored, how pages were generated on the server, and how everything needed to remain fast, stable, and secure.
AI helped tremendously throughout the process, but it did not eliminate the need to understand what was actually happening.
Whenever an error appeared or something did not work as expected, I still had to understand the underlying logic before I could fix it.
That was when I realized that designing a visual interface and taking a product all the way to something people can actually use are two different skills.
Once the core features were finished, another request came in: SEO optimization for every unique page, including individual news articles and faculty profiles.
Once again, it was something I had never formally learned. And once again, I had to learn it along the way.
Today, the website is live and reasonably stable. I am still certain there are plenty of things that could be improved. But one thing is clear: this was the first time I had taken a website from an initial idea all the way to something that other people could actually use—not simply designed the interface.
There has been plenty of criticism surrounding the term vibe coding, referring to the practice of building software with AI assistance without a formal programming background.
Some see it as a shortcut that cannot be taken seriously or responsibly.
I think it depends on how it is done.
As long as there is a genuine effort to understand the fundamentals and the logic behind each decision—not simply copying whatever the AI suggests—I believe the process can still be a valid way to learn and build.
What I Learned
- Taking a project all the way to something people can actually use is a different skill from simply designing it.
- Technical limitations, such as a hosting plan that does not support Node.js, can sometimes determine an entirely new learning path that you never originally planned to take.
- AI can accelerate the process, but it does not remove the need to understand the logic behind what you are building.