How to set AI guardrails in Quillio
To set AI guardrails in Quillio — open Firm Settings, AI Guardrails. Configure forbidden topics (e.g., do not give tax advice), required disclaimers on outputs, output length limits, and matter-type-specific rules. Changes apply to all matters in the firm.
Types of guardrails
Guardrails fall into three categories — topic restrictions (e.g., refer out tax or migration matters), output requirements (required disclaimers, required citation), and tone rules (e.g., never use absolute certainty on contested legal questions). Each can be configured independently.
Matter-type rules
Different matter types have different risk profiles. I recommend specific guardrails for high-risk areas — pro bono criminal matters, family law matters with risk factors, regulatory advice. Configure matter-type rules so they apply automatically.
Auditability
Every guardrail trigger is logged. If an output was modified because of a guardrail, the log shows the original and the modified output. This gives compliance officers visibility into AI behaviour.
Step-by-step
- Open AI Guardrails. Firm admin opens Firm Settings, AI Guardrails.
- Configure topic restrictions. Add forbidden topics. Quillio refers the user to appropriate resources rather than drafting on those topics.
- Set output requirements. Add required disclaimers, citation requirements, and length limits.
- Assign to matter types. Apply rules globally or to specific matter types. Test with a sample prompt to confirm expected behaviour.
Common issues
- Over-restrictive guardrails create user friction — tune based on feedback
- Disclaimers that are always added become invisible — make them purposeful
- Guardrails do not replace human review — they supplement it
Try Quillio on a real matter.
The fastest way to know if Quillio fits your practice is to use it on your own work. The free trial requires no credit card and no sales call.
Start your free trial