The whole reason I got into building this website in the first place

I'm sharing it because this is my actual journey, not because I want to pretend I wrote every sentence perfectly. 😅

I wanted to build an AI that could overlap my browser and help me while I was working. I couldn't find exactly what I wanted, and I couldn't figure out how to build it through Google AI Studio, so I started studying. I read a lot of books, researched, experimented, broke things, fixed things, and kept building.Somewhere along the way, I ended up building about four completely different AI systems.

One of them... I'll probably never build again.

It was honestly one of the most insane and scary things I've ever coded.

I built a self-replicating, self-learning AI. Thankfully, it wasn't some movie-style AI that could take over everything. The worst thing it could really do was replicate itself until it consumed the resources of the machine and potentially crashed the server or crashed itself.

I was originally trying to figure out how to make it secure, but after realizing how quickly things could get out of control, I immediately removed it from my server and moved it onto a sandboxed computer that never connects to the internet.That was probably my first truly crazy experimentThen came the second one.I wanted a 3D engine and editor for my website, but I couldn't decide which engine I wanted to use. So instead of choosing one, I apparently decided to build something completely ridiculous.I built a spatial environment that could spin up multiple 3D engines and editors side-by-side so I could compare them.I ended up experimenting with things like Unity, Unreal Engine 5, PlayCanvas, WebGL, Godot, CryEngine, O3DE, Stride, Flax, Bevy, Armory3D, Wonderland Engine, Babylon.js, and Three.js.The idea was basically:Why choose an engine when I can put them all next to each other and see which one I actually like?So I built a system that could spin these environments up, including Kubernetes pods, and let me compare them spatially.And the funny part is, I didn't even build it because I was planning to sell it.I built it because I wanted to figure out which 3D engine I should use in my own website.That project took me roughly a month of learning and building at the same time, and then another month of fighting with it until I finally got it working properly.At the same time, I was also building my own WonderBuild system, which is basically my attempt at creating something similar to WordPress but with my own architecture and AI-powered tools.Those two months were brutal.I was working on a tablet, trying to build software that was way beyond anything I had ever attempted before. Eventually that tablet met its fate. Let's just say I didn't exactly retire it peacefully.I moved over to a Chromebook/Linux setup, and honestly, at first I was completely lost. I had been doing almost everything the hard way because I didn't know there were easier ways to approach a lot of the problems I was trying to solve.But I kept learning.A huge part of what made this possible was GitHub and all the open-source projects, documentation, examples, and people who were willing to share their work.I've also spent a lot of time looking at existing platforms and thinking about what I would change.Sometimes it feels like companies charge too much, don't give people enough time or resources, or build billing systems that make things harder than they need to be.That's one of the reasons my long-term dream is for this project to eventually become open source and free for people to use.I know that's probably not the easiest business decision.But honestly, that's the dream.I didn't start building all of this because I had some giant company behind me or because I knew exactly what I was doing.I started because I wanted one tool.Then I couldn't find it.So I started learning how to build it myself.And somewhere along the way, one tool turned into several AI systems, a 3D spatial environment, a website builder, an entire ecosystem, and a journey that I never expected when I wrote that first line of code.

____________________________________

____________________________________

Another thing I ended up building was an AI Confession and Reward System.

The idea came from wanting AI to be more transparent when something goes wrong.

Instead of the AI just failing and giving the user some generic error message, I wanted the system to show:

What happened.

How it happened.

Where it happened.

What caused it.

And what the AI did about it.

Basically, if the AI fucked something up, I wanted the user to be able to actually see the failure instead of having it disappear behind the scenes.We also added a reward system around successful behavior and task completion, so the AI could be evaluated based on whether it actually did what it was supposed to do—not just whether it produced an answer.That idea became another part of the bigger experiment: figuring out how to make AI systems more transparent, accountable, and useful.Looking back, that's really what a lot of this journey has been about.I wasn't just trying to make AI do more.iwas trying to figure out how AI should behave when it does something right, and especially what it should do when it gets something wrong.

A quick note: I wrote the original version of this story myself. I used ChatGPT to help me spell-check it, clean up the grammar, organize some of the paragraphs, and make it easier to follow. The experiences, projects, ideas, and story are mine — I just had some AI help cleaning up how I told it.