Guardrails, by industry.
Use industry guardrails to define conversation boundaries for your business. Review the active rules, detection limits, and handoff settings before launch.
What guardrails are
Guardrails define intended conversation boundaries. Configured checks target recognized requests and responses outside the model. They can miss context or flag harmless content, so review and test the rules your business relies on.
How they work
Select the packs that match your business and review the active rules. Detected matches can trigger a configured action, including a refusal or request for team follow-up. Test the response and delivery workflow before launch.
Baseline rules for your business
- No discrimination or harassment
- No requests for sensitive personal data Fred has no business collecting
- Content grounding helps reduce unsupported answers; Fred is designed to acknowledge gaps
- AI identification should be configured and checked across the channels you use
Industry and regional guardrail packs
Industry packs address intended boundaries for healthcare, financial services, legal, insurance, real estate, home services, and other businesses. Configure Fred for general information and intake, reserve professional decisions for qualified people, and verify the active rules and tools.
Custom rules on top
Beyond the packs, every install can add custom guardrails: block specific topics, require disclosures, route sensitive subjects to a human, or customize the redirect language. Fred can even propose new guardrails from real conversation patterns for you to review.
An honest clarification
Guardrails are conversation boundaries, not regulatory certifications. HIPAA guardrails help you avoid inappropriate health conversations, they do not make your site HIPAA certified. Always consult qualified professionals for compliance in your industry.
Keep reading.
Find your industry.
Every vertical has its own guardrail pack and a page that shows what Fred can and will not do.