Building Trust into Insurance AI Native Products: Marsh Captive Solutions

Why AI is an amplifier of existing risk, and what that means for Chief Risk Officers in highly regulated industries. Governing it like financial or operational risk.Queena Cheung leads strategy and digital at Marsh Captive Solutions, where she built AI-native capabilities inside a heavily regulated business. ReadyCell compressed captive formation from weeks to minutes, earned the 2024 Business Insurance Innovation Award, and secured conditional pre-approved regulatory licensing from the DC Department of Insurance, Securities and Banking. Captive Companion replaced fragmented know-how with democratized know-what.In this conversation we cover:The gaps in traditional captive formation and management that made generative AI the right tool rather than a fashionable one.The hardest technical problem in the build: putting deterministic guardrails around probabilistic technology.Radical transparency with regulators, and why audit trails and human-in-the-loop architecture were the center of that conversation.How the UI itself communicates the AI and human boundary to clients.Role-based access control, sandboxing, evaluations, stress testing, and continuous monitoring as the concrete form of ethical principles.Where the trust but verify boundary is shifting toward human-on-the-loop and where it is holding firm.What separates captive programs using AI well from the ones still finding their footing: upskilling, culture, workflow redesign, and institutional intelligence.Why AI is an amplifier of existing risk, and what that means for Chief Risk Officers governing it like financial or operational risk.What Queena is watching over the next 12 to 18 months, including agent-to-agent interactions and the rise of AI employees.Her advice to any product leader building in a regulated industry: audit trails and data lineage before UI.About the GuestQueena Cheung is Chief Strategy and Digital Officer at Marsh Captive Solutions. She led the development of ReadyCell and Captive Companion and was shortlisted for the 2026 Captive Review Awards Emerging Talent recognition.About the HostPamela Gupta is the Founder of Trusted AI and the creator of the AI TIPS Framework, an enterprise AI governance architecture she developed in 2019. She is the 2025 recipient of the ISACA Joseph J. Wasserman Award for AI Governance and has chaired the GenAI Stage, presented on AI Cybersecurity and AI Governance globally and at the World AI Summit New York for six consecutive years. She holds CISSP, CISM, CSSLP, and AIGP certifications.Book NoteMy book, The AI Trust Layer: De-Risking AI Adoption, arrives September 1, 2026. It bridges the gap between identifying AI risk and operationalizing AI governance.LinksTrustworthy AI Briefing newsletter: https://www.trustedai.ai/briefing/Trusted AI: trustedai.aiConnect with Pamela: linkedin.com/in/buildingtrustedaiholisticallyCan Trustworthy AI help De-Risk adoption of AI? ‘Can Trustworthy AI can be  instrumental in helping organizations gain a competitive edge and  promote better business outcomes, including accelerated innovation with AI’.?With extensive experience in global industry leadership in areas of Business Strategy, Technology, and Cybersecurity, Pamela helps clients in creating a strategic approach to achieving business value with AI by adopting a holistic risk based approach to AI Trust. She defined 8 essential pillars of trustworthy AI.  Read more details at Trustedai.ai website.Her insights have shaped the way we look at the impact of Cyberwarfare on Business, strategies for efficient digital transformation, and governance views on Algorithmic failures.Join Pamela as she delves into her signature framework, AI TIPS, standing for Artificial Intelligence Trust, Integrity, Pillars and Sustainability. This podcast is all about operationalizing governance and building Trustworthy AI systems from the ground up. For questions or comments on this podcast reach out to me.