The Safety section in AI Agent Builder lets you configure:

• PII Masking: Detects and redacts sensitive data before processing or output.
• Guardrails: Controls what the AI Agent can process or respond to.

You can:

• Configure one PII Masking template per AI Agent

• Add multiple Guardrails

• Define workflow actions for Guardrail triggers

Add PII Masking

PII Masking automatically redacts sensitive information.

1. Open the required AI Agent in AI Agent Builder.

2. Click Define from the left pane.

   

3. Select Safety.

4. In the PII Masking section, click Add PII Masking.

   

5. Select a PII Masking template.

6. Review the PII Masking Name.

7. Click Save.

   

The selected template is applied to the AI Agent.

Modify PII Masking

• Click Change PII Masking
• Select a different template
• Click Save

Add Guardrails

Guardrails define what the AI Agent can and cannot respond to.

1. Open the required AI Agent.
2. Go to Define → Safety.
3. In the Guardrails section, click Add Guardrail.
4. Select one or more Guardrails from Select Guardrail.
5. Click Save.

Added Guardrails are displayed on the Safety screen.

Guardrail Scope

Guardrails can be applied to:

• Input messages
• Output messages
• Both input and output

Configure a Guardrail Action

By default, a block message is shown when a Guardrail is triggered. You can configure a workflow action for advanced handling.

1. On the Safety screen, expand a Guardrail.

   

2. Enable Include Guardrail Action.

3. Click Guardrail Action.

4. Configure the workflow in Dialogue Tree Builder.

   

5. Save the Dialogue Tree.

6. Save the Safety configuration.

   

   The configured workflow runs when the Guardrail is triggered.

Manage Configurations

From the Safety screen, you can:

• Change the PII Masking template
• Add or remove Guardrails
• Review Guardrail scope (input/output/both)
• Configure or update Guardrail actions
• Replace existing configurations

​