These are the most common ways to set up Averment. All of them use handlers so you have full control over what happens for each outcome.

Averment acts as an independent control layer for AI actions. It runs in real time, without human involvement, and gives consistent, predictable results.

Use this when an AI agent is taking actions autonomously. Instead of blocking and stopping, you let the agent respond to the outcome and adjust its behavior.

The agent does not just get blocked. It gets feedback it can act on.

Averment sends decisions to humans when needed, ensuring oversight for high-risk operations.

Use this when high-risk actions need a human to approve before anything happens. The action is paused and routed to an operator.

Use this when you want both automation and human oversight. For example, you might allow both allow and caution to proceed, use an AI feedback loop for review, and route block to an operator for review.

Use this when Averment sits at the center of multiple services or pipelines. Every system routes decisions through Averment and gets consistent enforcement everywhere.