Governing AI Agents

AI agents are moving beyond answering questions and increasingly taking action. As organizations begin putting them to work, new questions are emerging around accountability, control, security, cost and business value. At the CAC Industry Hub event on October 13, 2026, at the CBC Welcome Centre in Toronto, Karandeep Singh, Practice Director of CGI's Agentic AI and Data Practice, will speak on: Agent Governance: Scaling AI Agents with Control Karandeep will shed light on an urgent question enterprises face: When AI starts taking action rather than just answering questions, how do you keep humans accountable, maintain operational control and security, and measure and prove the business value? He'll share the practical Agent Governance approaches he's been developing and applying, which link an organization's operating model to the technical controls that make governance enforceable. He'll also cover the elements that make the approach work: Clear ownership: who owns each agent and makes decisions about it. A full lifecycle: from intake and approval through monitoring, renewal and retirement. Technical controls: agent identity, least-privilege access, human approvals, audit and monitoring. Cost optimization / FinOps: one view of agent costs, assigned to owners. Value realization: usage and cost evidence used to decide what to scale or retire. Support models: one front door, business champions, AI coaches and a builder center of excellence. The main message for leaders is that governance isn't meant to slow AI adoption down. It gives organizations the confidence and discipline to scale agents safely and transparently, with a clear link between risk, cost, adoption and business value. From the presentation to the conversation Karandeep's session will set the stage for an interactive peer discussion moderated by Tara Myshrall, Partner | BACG, Founder | Petra. Tara will bring the conversation into the room, inviting participants to share their experiences, questions and perspectives around issues such as: As AI agents take more actions, where should accountability sit? How are organizations deciding what agents should be allowed to access and do? Where should human oversight and approval remain part of the lifecycle? How do organizations balance governance, security and control with the need to innovate and scale? What are organizations learning as they move from experimentation to real-world deployment? There is no expectation that everyone around the table has the answers. Some organizations are already deploying AI agents, others are piloting them, and some are just beginning to explore what Agentic AI could mean for their business. Different stages. Different experiences. Different perspectives. That's what can make the conversation valuable. 28 people. Different experiences. One conversation. 📅 October 13, 2026 | 5:00–8:00 PM📍 CBC Welcome Centre, TorontoLight refreshments will be served.