In 2016 I admit I did not expect folks in AI to understand what I was doing and why. Much like today. But why? And What the… Brainf*ck is just eight simple commands (like >, <, +, -, ., ,, [, ]), it’s a Turing-complete and operates on an array of memory cells, essentially mimicking a stripped-down Turing machine. It’s was not meant for practical coding; it’s a brain-twister designed to challenge compilers and programmers alike, often used for fun, obfuscation, or proving concepts in extreme simplicity. Now, here’s where it gets exciting: I’ve pioneered using Brainfuck as the indigenous language (think native, foundational codebase) for agentic AI systems. Inspired by evolutionary biology it uses algorithms that “evolve” code through mutation, selection, and crossover to solve problems. By embedding Brainf*ck as the core “DNA” for these evolutions, I’ve created AI that generates and optimizes its own ultra-compact programs without the bloat of high-level languages. Why is this a milestone? It proves that even the most bare-bones syntax can birth sophisticated intelligence. Traditional programming often relies on verbose languages but Brainf*ck’s minimalism forces efficiency – every evolved instruction counts, reducing computational overhead and enabling faster iterations. It’s like evolving life from primordial soup instead of starting with complex organisms. The implications? This could revolutionize AI in resource-constrained environments (e.g., Robots or edge computing), make code evolution more accessible for hobbyists, and push boundaries in minimalist computing. Imagine self-improving AIs that write their own firmware in code that’s tiny yet powerful. I will write more about this and this time it won’t take me a decade like below. I also endeavor not to be as obscure so you can actually understand my stuff a little better. It will be an open source model that is not my primary model so that folks can experiment with it.