Axis Research
Axis is a research project exploring a universal semantic core for AI‑assisted programming: a tiny, explicit layer that captures the meaning of code, independent of surface syntax.
Semantic Core
A minimal meta-language: immutable values, pure functions, branching, composition. No loops. No mutation. No hidden runtime behaviour.
Function Registry
A directory of functions with stable identity and immutable semantics once published. When behaviour must change, publish a new function rather than rewriting history.
Bridge Layers
Real-world integration lives below the core: IO, concurrency, float semantics, OS/DB/GPU details. Bridges concentrate runtime complexity so the core stays small and analyzable.
Research First
This is not a product launch page. It is a living set of notes, experiments, and design constraints. Expect sketches, iterations, and careful language.
What we are trying to learn
- Can a tiny, stable meaning-layer reduce ambiguity for both humans and AI tools?
- What becomes easier when programs can be translated into one uniform representation?
- How far can immutability (of semantics) go before it fights real-world needs?
- Where should “messy reality” live, so the core stays clean?
Current status
Axis is early-stage research: much of what exists today is theory and sketches rather than finished tooling. The goal is to make the next steps concrete (a first compiler/interpreter, a minimal bridge, and real examples) and then adjust the design based on what survives contact with reality.
If you want to follow along, watch the GitHub repo for new specs and early implementations.