PMMs are increasingly being told that agents can research competitors, analyse customers, draft positioning, create launch content and support sales - removing the need to build a larger team. On paper, one experienced product marketer can now operate an entire function. In practice, that person becomes strategist, operator, editor, agent manager, quality-control department and organisational diplomat. The mistake is treating a team as a collection of tasks. Teams do more than produce deliverables. They challenge each other’s thinking, develop specialist expertise, preserve institutional knowledge, build trust across the company and share responsibility for outcomes. Agents can increase production capacity, but they cannot replicate those qualities. AI can, at best, imitate parts of the work, but it cannot replace the organisational system that makes the work effective. This session will examine the hidden labour behind the “one PMM plus agents” model and explain why more output does not necessarily create more impact (or indeed, good output). We will explore what happens when AI leverage becomes role compression: apparent productivity rises while judgment, creativity, learning and resilience deteriorate. Attendees will learn how to recognise this trap, think about what AI can actually do reliably, and design a PMM function that gains leverage from AI by using agents as additional capacity - not imaginary colleagues