You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
- Modified the demo to create a Guardrail with content filters,
topic denial, and PII redaction.
- Enhanced interactive mode for testing custom questions.
- Improved output messages for better user experience.
Generated by Copilot
Copy file name to clipboardExpand all lines: TestArena/Blog/AI/Bedrock/Guardrails/Index.razor
+1-1Lines changed: 1 addition & 1 deletion
Original file line number
Diff line number
Diff line change
@@ -67,7 +67,7 @@
67
67
<li><b>Multi-tenant systems:</b> Different guardrail versions per tenant means you can dial up or down the strictness for different customer tiers.</li>
68
68
</ul>
69
69
<CalloutBoxType="tip"Title="Guardrails are versioned">
70
-
<p>Because guardrails are published as numbered versions, you can safely iterate on your policy in staging while production stays locked to a verified version. No surprise policy changes mid-deploy.</p>
70
+
<p>Think of guardrail versions like Git commits — each version is a snapshot. When you publish version 2 with stricter rules, your production bot stays on version 1 until you explicitly tell it to upgrade. This prevents accidental policy changes from breaking production.</p>
0 commit comments