80s Computing Era Meets Small Local Decision Models

A developer has created 'amber-8', an assembly-like language interface that integrates small local LLMs to enable retro-style computing with modern decision-making capabilities. This approach splits a program into deterministic code and probabilistic human-judgment components executed by local models. This creative experiment bridges the gap between the analog world and digital programming by bringing advanced AI reasoning to low-level, retro computing architectures. It demonstrates the hacker spirit of combining old-school paradigms with cutting-edge local AI tools. The amber-8 project allows developers to write low-level instructions that interact directly with local models to evaluate conditions, parse inputs, and handle logic branching. Code snippets illustrate how text prompts can be translated into structured choices and scalar scores within an assembly environment.