Projects — research output
The Lab takes a single question — how do you actually use AI in real work — and answers it in three forms: a standard for how prompts should be written, a program for teaching that standard through real submissions, and an engine that enforces it automatically. Each one stands on its own; each one feeds the others.
Prompt Engineering Standards
A language specification for prose written to be executed. Undefined variables, ambiguous instructions, and hallucinations — treated as syntax errors. A ten-section standard that synthesises public guidance from OpenAI, Anthropic, and Google with patterns drawn from community practice.
Read the spec →Submit work, get it reviewed
A learning program designed as a single loop: lessons, submissions, reviews. Premium — human review for company teams, capped at five, currently full (an opening may appear in about three months). AI Review — automated review powered by CriticChain, in preparation. Curriculum and submissions are shared across both plans.
Read →Adversarial Review Engine
One AI adversarially audits the output of another. Eleven specialist agents run on LangGraph, looping through Draft → Critique → Refine until a review that merely sounds good has to either ground itself or be thrown out. An audit pipeline built to leave a trace — something defensible — in the age of AI governance.
Read →