How Gem Forge Idle Began

The first time I asked ChatGPT to help me make a game, I lasted about an hour. I had heard that it could write code, so I tried the most direct approach I could imagine: I gave it a prompt and expected a game to come out of the other side. It did not.

The attempt I abandoned

About a year before the real project began, I heard that ChatGPT could code. That immediately made me curious. I had an idea for a game, so I opened a conversation and asked it to make one for me.

For roughly an hour, I tried to turn that single request into something that worked. I did not really understand the tools, the workflow or what information the AI needed from me. The result was not what I had imagined, and I quickly reached the conclusion that game development was interesting but probably not for me.

So I stopped. There was no dramatic failure and no grand lesson at the time. I simply left the idea behind.

Why I tried again

Much later, the idea returned. I had noticed that many games displayed the Unity name when they opened, so I started looking into what Unity actually was. I discovered that I could install it and begin experimenting without paying for an expensive development setup. I asked ChatGPT whether it knew Unity and whether it could help me work inside it.

The answer was yes, but this time I did not expect one prompt to produce an entire game. I put ChatGPT in one window and Unity in another, and I started with something small.

I did not begin with a development pipeline. I began with ChatGPT on one side, Unity on the other and a lot of copying and pasting between them.

ChatGPT on one side, Unity on the other

The process was completely manual. I described what I wanted to happen. ChatGPT gave me code. I copied that code into Unity and tested it. When something failed, behaved strangely or needed to change, I copied the relevant code or error back into the conversation. ChatGPT changed it, I pasted the new version into Unity, and the cycle started again.

It was slow and repetitive, but it gave me something my first attempt never had: visible progress. Instead of asking for a finished game, I was solving one small problem at a time. Each successful change made the next question easier to ask.

From a moving sphere to a 2D game

The first things I saw were not gemstones, mines or polished menus. They were basic shapes: a small square, then a sphere. The sphere moved around, which was exciting for a moment because it meant that the instructions and code were doing something real inside Unity.

It also made the size of the challenge obvious. Movement was already becoming complicated, and 3D added another layer that I did not need for my first project. Rather than forcing the original direction, I changed it. I decided that the sphere did not need to move and that the game did not need to be 3D. I would build in 2D instead.

That decision mattered. It reduced the number of new problems I had to solve at once and gave the project a clearer shape. I was no longer trying to prove that I could make any kind of game. I was trying to make one specific, manageable game and learn what was required as I went.

Why an idle game

I enjoy idle games, so that became the natural direction. I understood the basic feeling I wanted: a game in which progress accumulates, systems open gradually and small improvements make a visible difference over time. I did not yet know how all of those systems would work, but I knew the genre well enough to recognise whether the result felt interesting to me.

The mining and gemstone project that eventually became Gem Forge Idle grew from there. It was not designed in one perfect conversation. It emerged through repeated questions, small tests, revisions and changes of mind. Some ideas stayed. Others were replaced as soon as I saw them working on screen and realised they were not right.

The beginning was not elegant, but it was real

Looking back, the most important difference between my two attempts was not a new piece of software or a perfect prompt. The second time, I accepted that the game would have to be built step by step. I stopped asking the AI to deliver the destination and began using it to help me take the next small step.

I still made every decision about what I wanted the game to become. ChatGPT translated many of those decisions into code, but I tested the results, explained what felt wrong and chose the next direction. The back-and-forth was the work.

What comes next in this story

This was only the starting point. During the first week, the empty Unity scene began turning into a recognisable game, and the copy-and-paste workflow produced both progress and plenty of problems. The next article in this development journal will follow that first week more closely, using the screenshots I preserved from the early build.

Gem Forge Idle did not begin because I suddenly knew how to make games. It began because, about a year after giving up, I was curious enough to try again—and patient enough to replace one impossible request with many small, answerable ones.