A two-axis decomposition of business AI work: deterministic versus semantic-judgment along one axis, pre-defined workflow versus exploratory along the other, yielding the Script, Algorithmic Search, LLM Workflow, and Autonomous Agentic Loop quadrants. These are not agent categories but problem-space categories that determine which architecture preserves accountability distribution. The quadrants form the diagnostic frame paired with the ten ADRs in AAP's two-axis structure.
Coined by Tatsuya Shimomoto (shimo4228) within the Agent Attribution Practice research line (introduced in the 2026-04-29 essay); two quadrant names were later renamed from the source essay for harness-neutral retrieval.
A two-axis decomposition of business AI work — deterministic vs semantic-judgment, pre-defined vs exploratory — yielding Script, Algorithmic Search, LLM Workflow, and Autonomous Agentic Loop. They are problem-space categories that determine which architecture preserves accountability distribution.
Tatsuya Shimomoto (shimo4228), in a 2026-04-29 essay from which the Agent Attribution Practice line was extracted (concept DOI 10.5281/zenodo.19652013).
業務 AI の仕事の 2 軸分解: 決定論 vs 意味判断の軸と、事前定義可能な workflow vs 探索的の軸。Script・Algorithmic Search・LLM Workflow・Autonomous Agentic Loop の 4 象限を生む。Agent のカテゴリではなく問題空間のカテゴリであり、accountability distribution を保つアーキテクチャを決定する。10 本の ADR と pair にして AAP の二軸構造の診断フレームを成す。