What are AI guardrails?
AI guardrails are rules set at the workflow template level that guide the Precisely AI assistant to follow your company’s policies when users interact with it in documents drafted from that template.
Guardrails help maintain compliance and accuracy across your organization’s AI-assisted workflows, and ensure that the AI assistant's answers align with your company's policies and guidelines.
How do I add guardrails to a workflow template?
Admins and Managers can create and edit guardrails within templates they have access to. The guardrails apply to all documents drafted from that template.
Simply,
-
Open a template.
-
Click Add guardrail.
-
In the text box, write a clear, specific rule that reflects your company’s policy.
-
Click Save.
Tip: For best results, ensure that you keep guardrails short, clear, and focused on a single policy point.
How do AI guardrails work in documents?
When users interact with the AI assistant in a document that includes active guardrails:
-
A green Active label is displayed below the AI assistant's text box.
-
Users can click the AI Guardrails link to view the full list of active rules.
-
Admins and Managers with template access can click Edit guardrails in template to make changes.
Any saved changes made to guardrails take effect immediately across all existing drafted documents, ensuring consistent AI behaviour and compliance.
How do AI guardrails influence responses from the AI assistant?
The AI assistant will provide responses in accordance with the guardrails defined in the template. For instance, it will avoid suggesting modifications that violate these guardrails and will evaluate the contract’s compliance based on the same standards.
Note: AI guardrails do not apply to external reviewers using the AI assistant.