Reality
Site, people, regulations, budget, climate and physical constraints.
AI can design the building. It cannot authorize reality.
REHT evaluates mandate, evidence, reversibility, semantic integrity and human authority before an AI-generated proposal becomes a consequential action.
Change the controls. The execution decision changes before the action does.
RACS decision
Composite risk
Mandate state
Evidence state
Receipt
AI-generated proposal
Unsigned demonstration receipt
receipt_version: architecture-demo-v2
action_id:
policy: architecture-execution-v1
decision:
next_gate:
digest_algorithm: SHA-256
context_digest: calculating…
receipt_version: architecture-demo-v2
action_id:
policy: architecture-execution-v1
decision:
next_gate:
digest_algorithm: SHA-256
context_digest: calculating…
The digest is real SHA-256. The receipt is not signed and is not production authorization evidence.
From requirement to physical consequence
The model generates options. The control system determines whether an option belongs inside the legitimate enterprise action space.
Representation
Original requirements, sources, ambiguity and transformations are preserved.
AI proposal
Models generate alternatives and simulations within declared scope.
REHT clearance
Mandate, admissibility, authority, risk and consequence are evaluated.
Governed execution
The action proceeds, changes, waits, escalates or stops with evidence.
Architects become more important not because AI lacks imagination, but because AI lacks mandate, responsibility and legitimate authority.
REHT governs the boundary where a generated proposal crosses into procurement, approval, construction or the physical world.