Back
Artificial Intelligence Underwriting Company
AIUC is building the trust infrastructure for enterprise AI — a SOC 2 equivalent for AI agents, combining a certification standard, red-team audits, and liability insurance tied to results.
Years
2025-2026
Role
Fractional Head of Design
Scope
Product Design, UX/UI, Prototyping, Visual Design
Challenge
Enterprise buyers evaluating AI vendors need confidence and to trust agents won't go haywire. But AIUC's evaluation data was dense, technical, and difficult to interpret without deep domain expertise — creating friction at exactly the moment a buyer needs to feel certain.
The challenge was designing a product that made rigorous AI safety and performance data legible to the people who actually make purchasing decisions, like CISOs and security professionals.
Solution
Framework-mapped information architecture Every result maps to a specific AIUC-1 requirement, giving buyers an audit trail from raw test data to certification standard — findings, not data points.
Trust through recency Models get updated. Buyers know this. Surfacing exactly when evaluations last ran addresses that skepticism without requiring explanation.
Multi-level navigation Evals generate enormous data. The dashboard lets executives stay at the summary level while giving technical reviewers a path to individual test cases, harm categories, and example prompts.
Legible test anatomy Each result surfaces the requirement, related real-world incident, harm type, attack tactic, and an example prompt — enough context to understand what was actually tested, not just whether it passed.
