My name is Azat. I am a software engineer who has recently become very interested in AI and LLMs, especially in coding.
Why this blog?
For the last seven months, I have been trying different approaches and techniques for agentic coding. There have been a lot of learnings along the way, and so far they have mostly lived in my head, which is honestly not the best storage.
Another reason is that I want to track both my own progress and the progress of the AI ecosystem. Since I am starting this blog after seven months of experiments, there will be some retrospective posts until I catch up to the present.
I am the main audience for this blog, and I will write posts for myself, on topics that I find interesting and worth remembering. But if you find my content useful, feel free to give feedback. I always enjoy fruitful discussions.
Topics
For now, the main topics will be learning AI — or vibe coding, context engineering, depending on what you call it — learning game development, and sometimes learning economics. I will explain in future posts why this combination makes sense to me.
Why AI
I do believe that AI — not necessarily LLMs — will have a bigger impact on humankind than the Industrial Revolution. The question is whether it will take years or decades. I am very excited to be living during this time, and I want to be at the forefront of it.
I have been keeping an eye on LLM development since OpenAI released ChatGPT to the public. It was very impressive, and I tested it regularly on different tasks. I kept thinking: “Having a smart chat assistant is awesome, of course, but what if it could write actual code?”
At that time, OpenAI had just introduced tools, so I wrote a small .NET application where I defined an interface with a single method and added a wrapper around the OpenAI API. You ask in chat what function you want, ChatGPT calls a tool that generates a class implementing that interface, the app hot-reloads it, and the new function becomes available.
It worked. ChatGPT was able to write real, compilable code, and the system could extend itself at runtime. In hindsight, this was vibe coding before the term existed — I just could not come up with anything more interesting to ask it to write than adding numbers or computing Fibonacci. So I got carried away with other topics.
Claude Code
After some months, real coding clients started to appear. Cursor was the first one I used. It was great, and it introduced me to the magical world of vibe coding.
But in August 2025, I had a “Claude Code moment”, and my life changed quite a lot. I became addicted to the dopamine this small CLI app was giving me. I cannot say that it was objectively better, but it felt better.
Suddenly, it could not only write code, but write compilable, “OK-ish” code. So I decided I had to build my own pet project using vibe coding.
The project
Now that I had a magic box — Claude Code — in my hands that could basically write anything, the question was: what should I write?
I have been playing computer games my whole life, and I always wanted to write a game myself. I even tried once with Unity, but after a month I gave up because of my laziness and the huge number of things I had to learn.
I have always wanted to write an economic simulator. My biggest inspirations were Capitalism 2 and Victoria 3. I thought: wouldn’t it be great to have a mixture of these two games, where you can be an entrepreneur or capitalist like in Capitalism, but with a sophisticated economic simulation like in Victoria?
So it was decided: I would vibe-code it.
To start, I needed to do thorough research. So for the next few weeks, I was “learning economics”.
It did not take long to realize that there was no way I could simulate a modern economy. It is just too complicated. So I decided that my game would take place during the Industrial Revolution.
First, it was a time when someone could become extremely rich in a short period of time. Second, it was a very interesting period, and there would be no lack of good stylized content for my game. And I also saw the bridge: the Industrial Revolution in the past, and the one that I believe is happening right now.
After that, I had very long and detailed conversations with ChatGPT, trying to understand how to model it.
I had only one requirement: money should not come from thin air and should not disappear.
That requirement alone made me understand that even the economy of that time was very complicated. For example, a factory pays a worker a wage. Some of it goes to buying bread, some to paying rent, and some to taxes. Then you start following the money: if a landlord receives a lot of money, where do they spend it? If the Crown collects taxes, where does that money go?
In the end, I came up with a long design document describing how my model should work.