My introduction to computers was when my dad got me a laptop when I was around 3. He used to bring cartoons, movies, and games from the office for me. Ever since then, I’ve had an obsession (sometimes unhealthy) with computers.
Fast forward a few years, we got a broadband service.
That was my introduction to the internet.
The internet opened new avenues and exposed me to ideas, I’d spend hours every day browsing the web, sometimes wasting time, sometimes learning new techniques but always having fun.
Eventually came the realization that instead of just passively browsing the web and using software, I could actually create my own.
This led me to research and learn a bit of HTML and CSS. I created my first webpage when I was 11. That was the beginning of a (still-ongoing) journey of just creating small scripts and programs.

A screenshot of the first web-page I ever made, in 2018.
I then learnt Scratch and published games, projects and even animations for everyone to see.
As time went by, I explored more realms: learning Python, creating automation scripts, learning a bit of Ethical hacking. I looked into anything and everything that I found interesting.
Stumbling upon a problem or getting an idea, figuring out how to execute it, writing every line of code from scratch (Okay, maybe some of it was copy-pasted from StackOverFlow), encountering problems, being frustrated and demotivated, spending long hours fixing it and then finally seeing your idea come to fruition.
While the final product was important, it was the whole journey that was rewarding and that I fell in love with.
A fair comparison would be an artisan who took pleasure in his craft; who spent years perfecting it and was proud of every product he made.
The AI Revolution:
Enter the AI Revolution. While AI is a broad term, for the scope of this article, ‘AI’ will specifically refer to Generative AI, Large Language Models (LLMs), and AI agents.
Coding is defined by Merriam-Webster as: the methods or process of creating or editing computer code. However, for the scope of this article, I would use what I call traditional coding (process of writing code by-hand) and coding interchangeably.
These LLMs quickly became extremely capable. They could answer queries, research, answer questions, understand nuances in language and most importantly: They could code.
And code, they could. Initially, their coding abilities were limited. They made basic logical and factual errors, hallucinated details out of thin air, and lost the context. A lot of these problems were solved as the models became more capable and with the introduction of agent harness.
In the starting (and as of the writing of this article), as a programmer, I personally only used it as a substitute for StackOverFlow - to look for documentation, to debug etc.
Even when I saw AI slop posts on LinkedIn related to Claude’s coding capabilities, I always scrolled past it.
At a certain point, it became unavoidable. There were friends who shipped capable software without knowing a single line of code; They made projects in hours that I could only imagine completing in months. And just to be clear, it also wrote code better than I could. While in the starting it looked like slop and the code itself was slop, experienced vibe coders made exceptionally well-polished products.
Therefore, it became clear to me that using AI agents for coding was inevitable.
On a side note: I will explore agentic AI in my upcoming blogs, so stay tuned.
This is similar to how the artisan would feel during the industrial revolution where suddenly a machine could make the same items but 10x cheaper and 50x faster.
They say programming was never about coding in the first place - that it was about solving problems and that writing syntax was only a tool.
Surprisingly, I mostly agree with that statement. With the exception of AI agents that autonomously make design decisions, code, solve problems - Clearly a lot more than simply writing the code. For those cases, an input is a prompt and the output is the finalized product. It might take a few extra prompts to get the final product to the human’s liking, but that’s it. Even then, part of the arguments would still remain valid so long as a human is involved at any stage.
They say, humans would still act as managers of AI and software engineering jobs would just shift in that direction (Perhaps a better name would be ‘software managing’?). So, to be a better manager one would still need to be a better software engineer.
Anecdotally as well, I’ve talked to multiple capable seniors in academia and software-engineering jobs who are encouraged to utilize AI in their workflows. It does seem that previous development skills do correlate positively with better outcomes even with AI.
But 6th-grader Yusuf, who fell in love with programming and decided to pursue it as a career, didn’t sign up to be a manager or a prompt engineer.
He signed up for that joy of building something new, to fail and figure it out, to spend hours trying to fix a bug only to find a tiny typo being the problem.
Well, traditional coding is dead.
So is the kid that found pleasure in it.
Traditional coding is replaced with Vibe-coding.
The kid is replaced with a 19-year-old who realizes he needs to adapt.
As with all things, he will move on, as he always does.
But let him mourn.
At least for a bit.