AI agents are beginning to move from advice and drafting into the actual flow of work: resolving tickets, preparing correspondence, routing approvals, and initiating next steps. As more work is carried across systems with less step-by-step review, accountability becomes harder to locate and oversight can become performative if it is not designed carefully. The real governance problem is whether leaders can trace the handoff, explain the decision, and intervene before speed becomes unmanaged risk.
This Session Will Examine:
- Defining who owns the decision when an agent acts: approval authority, escalation triggers, and where a human signature is still required.
- Oversight that works in practice: action logs, audit trails, exception review, and thresholds that route higher-risk actions back to a person.
- The operating model among HR, IT, legal, risk, and business owners: who grants access, who monitors activity, and who can shut an agent off.
- Disclosure norms for employees, customers, and counterparties when work is agent-produced.
- Proving value when output is co-produced: measuring quality, error rates, and cycle time against the human baseline.
Interested? Add to Calendar
