16 July 2026 · KYE Protocol™ · Doctrine · ― Is Not Authority

Observability Is Not Authority.

You can watch an AI agent take an action in perfect detail and still have no answer to the only question that matters at the boundary: was it allowed to?

See Authority Finality™More doctrine

Observability Is Not Authority, point by point

Demo data

You learn: The page's points, one at a time. You can: Step through them.

  1. Step 1 of 3

    KYE Protocol™

    Live metrics on latency, cost, tool calls, and token spend.

  2. Step 2 of 3

    KYE Protocol™

    Drift and quality signals on model outputs over time.

  3. Step 3 of 3

    KYE Protocol™

    Alerts when behaviour leaves a known baseline.

How this widget is built
Demonstrates
KYE Protocol™
Components
  • The points this page makes
Flow
Each point in order, as the page sets it out. Architecture diagram
Used on

Watching an agent act is not the same as deciding it was allowed to act.

What observability gives you

Observability answers "what is the agent doing?" — and it answers it well.

  • Live metrics on latency, cost, tool calls, and token spend.
  • Drift and quality signals on model outputs over time.
  • Alerts when behaviour leaves a known baseline.

That telemetry is real and useful. It is a lens on behaviour. But a lens reports; it never decides.

Where it stops

Observability is downstream of the action. By the time it sees the event, the agent has already acted.

Monitoring tells you an agent moved a payment. It does not tell you whether that agent held delegated authority to move that payment, for that principal, within that purpose, at that moment. Those are authority questions, and no dashboard resolves them — a chart of what happened is silent on what was permitted.

  • It observes actions; it never admits or denies them.
  • It sees an outcome; it holds no record of the grant behind it.
  • It is after-the-fact by construction — the action already shipped.

The line

Authority is decided before the action, at the boundary, and sealed into evidence.

KYE Protocol™ checks a request against the delegated authority the principal actually holds — purpose, scope, and expiry — and returns an admit-or-deny verdict the instant before the action, then binds that verdict into a replay-verifiable Evidence Pack™. That is Authority Finality™: the right to act is settled at the boundary, not narrated afterward. Observability watches the run; Authority Finality™ decides whether the run may happen.

Honest scope

KYE Protocol™ proves authority, not truth.

It does not claim the agent's answer was correct or its model well-calibrated — that is the work of evaluation and observability, and both remain necessary. KYE Protocol™ proves exactly one thing, and proves it completely: whether the actor was authorised to take the action. Keep your telemetry. Add the layer that decides.

Watching an agent act is not the same as deciding it was allowed to.