Almost every procurement team has AI running somewhere. Very few have managed to scale it.

Most organizations have landed in the same place: automated workflows with a human checking every output. That is real progress, and it is where the majority stall. Scaling is not a matter of adding more agents. It is a different problem with a different answer.

This session is a candid conversation for teams ready to move past that plateau. Our panel discusses why individual task-specific agents rarely scale, and why what scales is a governed workflow rather than a standalone agent.

From there, the discussion turns practical. The panel shares a six-part test they use to decide where AI should be deployed first, separating a use case that will scale from one that stalls after the pilot, and works through real examples of agentic AI running in procurement today.

The conversation closes on governing it: where a human should stay in the loop, what guardrails need to exist before an agent acts on its own, and the signals that tell you when it is safe to widen the boundaries. Plus the step most teams skip, which is what to baseline before you deploy, so results can be attributed and the next phase can actually be funded.