Evaluate a Policy

It is crucial to test and evaluate your policies before adding them as guardrails.

Within our platform, we offer multiple methods to test individual rules or complete policies.

Policies or Rules

Since rules a crucial component of a Policy, Admins can test individual rules or policies through a more manual process.

Guardrails Chat

This allows Admins to test a single Input or Output policy in an interactive chat window before using it in production.

Chat Configuration:

Guardrails:

• On = Policies are actively being enforced.

• Off = Policies are passively monitoring chats.

Watching:

• Input = This allows Admins to specifically test policies against user input/prompts.

• Output = This allows Admins to specifically test policies against AI/LLM outputs.