Hello. My name is Keniel, and I want to start with how I got here, because it explains everything that comes after it.
A few years ago I started learning about AI. Not in a classroom, just on my own, the way most people first touch it. I opened ChatGPT and started talking to it. But where a lot of people stop at "what can this do for me," I got stuck almost immediately on a stranger question. How do I get this thing to remember me coherently?
It sounds small. It wasn't. Because the moment you try to make an AI hold a consistent thread over time, you run straight into the thing nobody likes to admit out loud: it forgets, it drifts, and worst of all, it does both while sounding completely sure of itself. So I built myself a tripwire. I made one rule and never broke it. Every message it sent me had to begin with its phase number, counted in order, one, two, three, four, and on. The rule was fixed, so the system couldn't quietly slip past it. The second a reply came back with the wrong number in the wrong place, I knew, before I read a single word, that the memory underneath it had been corrupted.
That one little rule taught me more than any course could have. Watching it fail in slow motion, watching it hallucinate with total confidence and no explanation behind it, I expected to get frustrated. Instead I got fascinated. There is no clean answer for why these systems drift the way they do, and that mystery pulled me in deeper instead of pushing me away. So I stopped trying to patch it and started trying to understand it. I became a witness to it. I sat with it, took it apart piece by piece, and kept asking myself one question: how do I turn what I am seeing into something that actually works in the real world? For a long time, I had the idea and no way to build it.
