Skip to main content
After an assessment is submitted, the assistant reads your answers, the risks that were found and the conversation, and suggests settings for the guardrails in scope. A suggestion is never applied for you: it goes into your draft, where you review, test and apply it. Suggestions cover the risks the assessment found and the risks Altrum applies to every AI system. Risks you added yourself are yours to set up.

​
Add the suggestions to your draft

Open the guardrail editor for the AI system. While the suggestions are being prepared, a banner says “Your assessment is being used to suggest guardrails.” When they are ready:
1

Read the banner

The banner Your assessment suggests settings for your guardrails says what adding them does:
  • “Switches off, because your answers say it does not apply:” names guardrails the assistant thinks you do not need.
  • “Leaves for you to set up, the assessment did not give enough detail:” names guardrails you need to fill in yourself.
2

View them

View suggestions shows each suggested setting, read only.
3

Add them

Add them to my draft replaces your working draft for the guardrails the suggestions cover. The message says “Suggestions are in your draft. Review them, then apply.”
4

Finish the draft

Review every guardrail, set up those left for you, and Save draft. Then test, and apply or complete the assessment. Nothing is enforced until you do.
A guardrail the suggestions switched off still shows as Needs your action in the assessment results, so you confirm the decision yourself.

​
If suggestions could not be prepared

The banner says “Suggestions could not be prepared from your assessment. You can still set guardrails up yourself.” Choose Try again, or configure the guardrails by hand. See Configure guardrails.