AI agents are rapidly moving from recommending what customers should buy to taking action on their behalf—searching products, comparing options, applying offers, interacting with commerce APIs, and potentially completing transactions. That shift changes a fundamental assumption of digital commerce: the person making the decision may no longer be the system making the request.

Your Agent Has a Credit Card Now explores how zero-trust principles can be applied to agentic commerce when autonomous agents interact with high-value enterprise systems. Rather than implicitly trusting an agent because it represents an authenticated customer, commerce platforms must continuously verify identity, intent, permissions, context, and risk before allowing an action to proceed.

Through a practical agentic commerce journey, this session examines patterns for delegated authorization, least-privilege tool access, scoped credentials, transaction limits, step-up approvals, human-in-the-loop controls, observability, and auditable agent actions. It also explores failure scenarios—from manipulated agent instructions and excessive permissions to unintended purchases—and how architecture can contain their impact.

Attendees will leave with a practical framework for designing commerce experiences where AI agents can discover, decide, and act without giving them unlimited authority—enabling the next generation of autonomous commerce while keeping trust, control, and accountability at the center.