The Stepped L7-to-L0 Hierarchy

Command authority descends in steps from strategy at L7 to microsecond actuation at L1, tightening latency and narrowing scope at each step.

Authority flows downward

While telemetry rises from the plant, authority descends. The blueprint is often drawn as a staircase from L7 at the top to L0 at the bottom, each tread handing a narrower, faster, more constrained decision to the layer below it.

The descent

• L7 Ecosystem sets plant-level intent — grid dispatch windows, maintenance schedules, regulatory constraints.
• L6 Experience presents that intent to operators and captures their confirmations.
• L5 Copilots translate intent into engineering goals — a target equilibrium, a plug-density setpoint.
• L4 Orchestration routes goals through the rules engine and safety bounds before anything reaches control.
• L3 Twin & AI computes the actuation trajectory using the predictive shadow and MPC agents.
• L1 Control Plane executes the trajectory deterministically and holds the autonomous failsafe.

Each step tightens the contract

As commands descend, three things tighten simultaneously: latency budgets shrink from minutes at L7 to microseconds at L1; scope narrows from plant strategy to a single actuator; and the ability of AI to intervene falls to zero at the bottom, where the hardware failsafe is purely physical. This is intentional — the most consequential, least reversible actions are the least dependent on intelligence.

Why steps, not a single jump

A direct link from strategy to actuator would couple a slow, uncertain decision to an irreversible one. The stepped hierarchy inserts validation, human oversight, and safety bounds between intent and action, so that no high-level goal can bypass the checks that protect the breeder's superconducting magnets or the burner's high-field plug.

The reverse path — how raw diagnostics become decisions on the way up — is described in from diagnostics to decision.