---
title: "Threshold escalation · agent human oversight · Agent Fieldbook"
url: https://agentfieldbook.org/security/oversight/threshold-escalation/
description: "The agent resolves routine cases autonomously and escalates low-confidence or high-risk ones to a human. A confidence or risk score decides the route."
section: "Security · Human Oversight"
source: Agent Fieldbook — generated from the published page
---

# Threshold escalation

The agent resolves routine cases autonomously and escalates low-confidence or high-risk ones to a human. A confidence or risk score decides the route.

Verified against [nist.gov](https://www.nist.gov/itl/ai-risk-management-framework)

- [nist.gov](https://www.nist.gov/itl/ai-risk-management-framework)

## How it works

## At a glance

| Field | Value |
| --- | --- |
| Mode | HOTL |
| Exemplifies | support |
| Example | Support pods· fraud-review queues· tiered resolution agents |

## Basis

## Verified against

- [https://www.nist.gov/itl/ai-risk-management-framework](https://www.nist.gov/itl/ai-risk-management-framework)

The corpus does not rewrite vendor documentation. Where copy is quoted verbatim it is marked as such and attributed to its source.