Execution is the control point
Most sustainability software operates after execution. It measures impact, explains historical usage, or recommends better placement. That is useful for reporting, but it is not governance.
Pre-execution governance moves the control point upstream. A workload asks to run. The authority layer evaluates the request against environmental signals, policy constraints, and execution posture before compute is admitted.
Authorization changes the category
The critical distinction is whether the system can bind the outcome. A reporting tool can describe what happened. A scheduler can recommend a cleaner region. A control plane can return one of several binding actions such as run, reroute, delay, throttle, or deny.
That changes the product from advisory software into operational infrastructure. The system does not describe behavior at the edge of execution. It decides whether behavior is allowed at all.
Environmental governance is multi-objective
Carbon is not the only signal that matters. Water stress, latency protection, and operating policy all shape whether a decision is defensible. A real governance layer must combine them without hiding the trade-offs inside black-box heuristics.
CO2 Router uses SAIQ governance to evaluate those constraints before execution. The result is attached to the decision frame so trace, replay, and provenance remain consistent with the decision that was actually enforced.
Proof is part of the contract
Pre-execution governance only matters if the resulting decision can be inspected later. That requires proof, trace, replay, and provenance to stay attached to the same frame, rather than being reconstructed later from best-effort logs.
The system therefore needs deterministic replay, trace-backed decision state, and verified environmental inputs. Without that, governance is only a narrative.