How AI changed my job

Introduction

"This article is primarily addressed to people who are not very technical or individuals working in IT who are having their first experiences with the new Large Language Models (LLMs)."

I would like to start by explaining the reason for this article: we are in a time of great and rapid change, with a lot of misinformation and distrust regarding the use of AI tools. Literally, something new comes out every week, and this can be really disorienting for most people. While I agree that AI tools should be used with caution and awareness, I am convinced that they are the beginning of a new industrial revolution that will completely change the way we work. Like all tools, they can be used well or poorly and have good or bad results. In my opinion, AI is not replacing developers, but it is radically changing the way we should approach software development.

My background as a full-stack developer with a strong focus on technology and code optimization.

"The traditional approach: Spending significant time on writing, optimizing code, and learning best practices."

I started developing software at a very young age and spent countless hours studying, understanding, and experimenting. I began when the only tools were huge (and expensive) books where you had to literally read all the details if you didn't want to miss any important concepts. My classic approach was simple (as I think everyone else in the field): study for days, experiment, and then apply. Training courses and exchanging ideas with colleagues. Every step forward added new knowledge to the growing pool of information.

If I had to give a simple summary of my journey as a programmer (excluding my background in operating systems and networks): C, Perl, Visual Basic, HTML, JavaScript, Java, C#, Objective-C, Swift, Python, and many, many more. Not to mention the study of all the libraries, patterns, and frameworks. It was continuous learning that I fortunately loved and found really fun and stimulating. Thanks to the internet, access to information about languages became easier and it was no longer necessary to memorize all the information to be efficient. However, the continuous search for the most elegant and efficient ways to write code was added to the workload. It was a journey that I really appreciated but sometimes (especially with less interesting projects) it was really exhausting.

"The paradigm shift: How AI has fundamentally altered my workflow and focus."

I remember the early code correction tools from JetBrain and the first version of Copilot (from Github) that recognized code patterns for correction and optimization. Even back then, they provided significant help (we're talking about 4 or 5 years ago), but with the arrival of Large Language Models, the world of programming, my world, has completely changed. While I'm writing code, 9 times out of 10, the editor automatically completes what I'm writing correctly or partially correctly, but generally in the direction I had in mind. Currently, the two editors I use the most are Cursor and Windsurf. Not only are they great at suggesting, but they can also create programs from scratch. They are not perfect, of course, often producing stereotypical code based on the input used for the model's training. But they are an excellent starting point.

The routine now is: starting from a template provided directly by the package manager (like NPM), we create the base of the program and then ask the AI to add the required features. In a few minutes, we achieve the same result that would have required a lot of study and experience. At this point, given the basic structure of the technology and the added features, we can focus on corrections and optimization in the direction we need. Some people tell me, "yes, but the programs are all the same," but actually, no, because it heavily depends on the prompt (direction) we gave the model at the time of the code generation request. Obviously, the more detailed the context and request, the closer the produced code will be to what we expect.

"AI as a brainstorming partner: Exploring new concepts and possibilities."

It's like having a colleague who proposes solutions, with whom you can discuss the best approach. You can ask questions about the written code, get explanations, advice, and most importantly, direction. Using AI has given me a sense of liberation; I am free to think about things that I would have previously discarded due to time (and therefore cost) constraints. The only limit at this moment is imagination.

My programming work has shifted from writing code to thinking about what might be the best solution regardless of my technical knowledge. If I need to create something with tools I am not familiar with, I ask for the generation, analyze the code, and have the AI explain the reasoning behind certain choices and if there are possibilities for optimization.

Before, I used to write with the idea that everything had to be optimized from the start (especially for prototypes), but now my goals have changed. I can produce more things, faster, and focus on refinement and optimization only when it is truly necessary.

I could summarize by saying that AI is giving me the ability to:

• Fast-track understanding of new technologies and their potential uses.
• Interactive Q&A with AI to deepen knowledge and clarify concepts.
• Shifting focus from code details to higher-level strategy:

Finally, I can focus on the bigger picture, optimizing ideas and the overall process.

We are finally shifting from code-centric to idea-centric development to pursue faster innovation and much more efficiency.