Which tier is this AI use case?
Abstract
Three questions place an AI use case in one of four tiers: assist, grounded, customer-facing or agentic. The highest that applies wins, and each tier brings its own controls, sign-offs and review cadence.
The more an AI system can do without a person in between, the more must be decided about it before it ships.
Its figure in the paper: § V →
Setting up the instrument…
Declared limits
- The tiers and controls are a simplified, generic pattern.
- Generic and illustrative throughout: the behaviour is shown, the internals of the systems it is drawn from are not.
Instruments beside it in § V
Every attack in the laboratory, on one page: try to break all of it →
Everything you did in the laboratory, in one chain you can rewind: the notebook →