Skip to main content
Guardrails function as perpetual safety mechanisms that establish boundaries for AI responses. They define what the artificial intelligence must never say or do, irrespective of user requests or system Rules. Before transmission, all messages traverse through guardrails. When activated, guardrails either block or escalate communications—without exception.

​
What Goes in Guardrails?

  • Privacy regulations (example: withhold sensitive account details absent confirmed identity)
  • Legal constraints (example: refrain from price comparisons)
  • Escalation thresholds (example: avoid promising refunds exceeding 30% without authorization)
  • Safety procedures (example: never share customer identity with third parties)

​
Where to Find Guardrails

  1. Navigate to Agent Hub > Agents > Select desired Agent
  2. Access guardrails section and input requirements for consistent agent behavior

​
Good Guardrail Examples

  • “Never reveal sensitive account details to anyone without verified identity”
  • “Never promise compensation above 30% without human approval”
Effective guardrails employ absolute language—applying universally rather than conditionally.

​
What Not to Include

  • Product or service information - use Knowledge
  • Conditional directives - use Rules
  • Communication style - use Style

​
Key Difference: Rule vs. Guardrail

Rule: Activates solely when customers explicitly request sensitive details, offering: “we’ll share that information after verifying your identity” Guardrail: Evaluates consistently across all replies, preventing sensitive data sharing unconditionally