Notes from the shadow.
System Zero, what we measured, and how to run a reflex in front of your LLM. Losses included. RSS
2026-10-09 · 5 min read
Zero-code System Zero for the OpenAI Decisions APIPut a certified reflex in front of client.decisions.create with three commands and no code change. Shadow mode first, the dashboard, the agreement check, the cost maths and the limits.Read →2026-10-09 · 3 min read
Snake at 120 moves a second: System 0 vs System One models, at game speedshad0w learned snake and Pong from an LLM's answers. Then it raced three System One decision models on an Apple M5 Pro's GPU and a hosted LLM through the same games, with every decision timed. Here is what happened, and the Pong you can play.Read →2026-10-09 · 5 min read
94.3%, 51.3%, 0%: honest numbers from shadowing a hosted LLMWe put shad0w in front of a hosted LLM through the official OpenAI client and measured how much shad0w answered, how often it agreed with the LLM, how fast, and where it answered nothing at all.Read →2026-10-08 · 2 min read
Adding a category without breaking productionA trained shortcut and a changing list of options are a dangerous mix. The rule shad0w follows, and what happens when you add, remove or rename an option.Read →2026-10-08 · 4 min read
System Zero: the reflex layer below your LLMshad0w is the shadow of your LLM. It watches your LLM, learns to answer like it, and takes the repeat traffic, so the LLM only gets what shad0w is not sure about. What System Zero is, why a cache is not enough, and where shad0w stays silent.Read →2026-10-07 · 5 min read
How shad0w proves it agrees with your LLMThe certificate behind shad0w, in plain words: a held-out test, a confidence line, an allowance for bad luck, and what the promise does and does not cover.Read →