Organizations run on decisions, yet decisions are the one thing they never record, observe, or evaluate. A framework for instrumenting, evaluating, and improving organizational decision intelligence — grounded in bounded rationality and sequential decision theory, and built to be implementable in any organization.

An LLM agent is the ultimate confused deputy. The Agent Proxy Pattern gives it a scoped, capped, revocable mandate instead of raw credentials — every action authorized, settled, and audited — and, because loss is provably bounded, makes agent actions insurable.