Approval, queue and block rates
How much of the lane still clears automatically, and what proportion already requires review to stay safe?
NORNR
A cold reporting frame for what agent spend teams should measure before they trust a lane.
NORNR / State of agent spend
Category assetThe first report should define the measurement frame, not bluff precision the market does not have yet.
This edition is intentionally cold. It defines the metrics, failure classes and export expectations that decide whether a lane is merely autonomous or actually governable.
What to measure
Spend is not enough. The posture around the spend is what determines whether a lane is safe to widen.
These are the category metrics NORNR cares about first.
Receipt completeness and export readiness
If the lane spends but cannot export a calm packet later, it is not production-ready in finance or risk terms.
Velocity drift and first-seen counterparties
These are often the earliest signs that a lane widened faster than the mandate did.
Review burden and exception burden
A lane that is technically safe but operationally unreadable still fails the institutional test.
Finance packet completion and reconciliation posture
The month-end view matters because many lanes look fine until the close bundle has to survive outside the control room.
Destination and vendor posture changes
Production maturity often fails not on budget amount but on unexplained destination drift.
Failure classes
The market should expect these classes to show up early.
A report becomes useful when it names recurring failure modes and ties them to the control surface that neutralizes them.
Retry loops and silent provider bursts
Most model-spend lanes fail through repetition before they fail through one obviously catastrophic action.
Paid-step fan-out and final-click ambiguity
Browser lanes fail when the expensive moment is hidden inside many harmless-looking steps.
Local tool authority without institutional review
Desktop and local-agent lanes fail when teams mistake capability exposure for control.
Evidence gaps discovered only at close
A lane often looks safe operationally and only reveals weakness when finance asks for one defended record.