How do enterprises safely deploy AI agents when those agents can access data, call tools, trigger workflows and make consequential decisions?

In this episode of the Chief AI Officer Podcast, Sanjay Puri speaks with Jamie Campbell, Head of AI at Manulife Canada, about what it takes to build responsible, governed and scalable AI in a highly regulated enterprise.

Jamie explains why AI governance must be treated as an operating system for AI — not an afterthought. He shares how Manulife approaches AI risk, model evaluation, human-in-the-loop decision-making, agentic AI safeguards, enterprise access controls and the challenge of deploying AI across multiple jurisdictions.

The conversation explores:

  • Why AI governance needs to be built into AI systems from day one
  • How Manulife’s automated underwriting engine uses AI to support life insurance decisions
  • Why prompts alone cannot be trusted as AI safeguards
  • How enterprises can evaluate frontier AI models before putting them into production
  • The control mechanisms needed when AI agents access data or trigger workflows
  • Why enterprise-level permissions and access controls are critical for agentic AI
  • How to manage AI risk across the technology supply chain
  • Build vs. buy decisions for enterprise AI platforms
  • How to balance global AI architectures with local regulatory requirements
  • Why AI champions and change management are essential for adoption
  • How AI leaders should measure ROI through tangible business outcomes
  • Why responsible AI sometimes means deciding where AI should not be used

Jamie also shares a practical framework for AI governance, responsible AI, model evaluation and agentic AI deployment that Chief AI Officers, CIOs, CTOs and enterprise AI leaders can apply to their own organizations.

Key themes: AI Governance, Agentic AI, Responsible AI, Enterprise AI, AI Risk Management, AI Regulation, AI Agents, AI Security, AI Compliance, AI Model Evaluation, Human-in-the-Loop AI, AI ROI, Frontier Models, Insurance AI, AI Transformation.

Chapters:

00:00 Building Trust in Enterprise AI

01:33 What Is Manulife?

02:07 AI in Highly Regulated Industries

03:34 Governance as the Operating System for AI

04:43 Why You Can’t Just Trust the Prompt

05:56 AI-Powered Automated Underwriting

07:03 Human-in-the-Loop AI Governance

10:06 The One Control Every Automated AI System Needs

11:31 Build vs. Buy for Agentic AI

14:05 Managing Frontier Models and AI Supply Chains

17:27 Evaluating AI Models Before Production

19:08 Controlling What AI Agents Can Access

20:44 Global AI Models vs. Local Regulation

22:44 Building an AI Champions Network

24:33 AI Adoption, Change Management & Leadership

27:29 Measuring AI ROI and Business Impact

29:23 Lightning 

Author

About the Author: Author

Chief AI Officer at Ameritas

Leading enterprise AI strategy and technology integration. Advises Fortune 500 boards on high-value generative AI deployment architectures and governance frameworks.