Purposing · P-GATE · AI assistant
AI Gate — the readiness verdict
Purposing does not end when the Purpose is written — it ends when the engagement passes a readiness verdict. The gate is a disciplined pause that asks: is this ready to build, or would we be building on sand? It runs ten checks over the whole stage and returns GO, GO with owned conditions, or STOP. Passing it is what earns the right to start scenario work. A condition is not real unless it has an owner and a by-when.
The ten readiness checks
Purpose is singular · Define
one purpose, one sentence
Purpose is contextual · Define
written on the underlying need, not the ask
User and use are clear · Define
one named decision-maker, a real dated decision
Purpose is learning-framed · Define
test / prepare / explore, not a target
Horizon is set · Design
the span is named and workable
Threads are contextual · Diagnose
the critical uncertainties are the right ones
Sensitivities are managed · Contract
any high-sensitivity assumption touching a national target has an owned, dated pre-brief
Timing is workable · Design
the windtunnel and use event do not collide (e.g. budget season)
Re-perception is reachable · Contract
the shift the work must achieve is achievable and pre-briefed
Assumptions are surfaced · Diagnose
the register exists and is owned
With an ANTHROPIC_API_KEY configured this calls Claude (claude-opus-4-8) as a sceptical gatekeeper; without a key a transparent demo ruleset runs. The verdict is a facilitator's aid — the gate is a decision made in the room, not by the model.