Show HN: 80s computing era meets small local decision models (github.com)

Amber-8 introduces a novel small deterministic stack machine that transforms messy, subjective data into structured outputs by leveraging a local language model, reminiscent of early computing principles. This innovative system counters the complexities of modern programming by using only three straightforward instructions to interface with the model, focusing on clarity and predictability. For instance, Amber-8 can process vague feedback about room comfort, allowing for systematic adjustments based on user responses without the fragility associated with regex or autonomous agent frameworks. This development is significant for the AI/ML community as it emphasizes a return to simplicity and determinism in programming, aiming to enhance accountability and auditability. By splitting the process into two distinct layers—judgment, handled by the language model, and procedural execution in clean, deterministic code—Amber-8 ensures that programs remain easy to debug and trace. With compiled binaries under 2 KB and designed to run on minimal hardware, this micro-computing approach not only retains the elegance of earlier eras but also provides a robust framework for incorporating machine learning into practical applications, eschewing the unpredictability of current trendier models.