#Architecture

May 30, 2026

A Thin LLM Provider Layer — Just Thin Enough To Swap

I route every model call through one small abstraction. It's deliberately tiny, and that's the whole point.