Building Recourse.ai: A Parametric Insurance Prototype for the Age of Agentic AI

Short summary

Recourse.ai is a prototype applying parametric insurance principles to AI agent risk — payouts trigger automatically when agent error rates or guardrail breaches cross pre-agreed thresholds. The approach replaces static E&O paperwork with live operational telemetry, creating incentives for better guardrails. The author built an Android prototype targeting the Lloyd's Lab ecosystem.

• •Parametric insurance model triggers automatic payouts when AI agent safety thresholds are breached
• •Guardrail Maturity Score quantifies risk from live agent behavior rather than self-reported checklists
• •Prototype modeled payouts clearing in under 48 hours vs 90+ days for traditional E&O claims

Generated with AI, which can make mistakes.