An engineering manager wants to add an agent to the team. Nobody owns the answer. Onboarding AI teammates is org design, and HR is the department that already knows how teams work.
One construction RFI, twelve days in queues, ninety minutes of actual thinking. The queue is the problem, not the people, and AI agents are quietly attacking the waiting.
As agents absorb status archaeology, first drafts, and transport meetings, the PM job shrinks back to judgment and taste, plus one new skill: designing workflows for teams that are only partly human.
The writing takes three days. The piece ships in five weeks. Content velocity was never a writing-speed problem, and the teams getting faster are redesigning handoffs, not buying a faster pen.
Two companies, one identical payment webhook failure, about the same engineering time. One incident ran 54 minutes, the other half the night. The difference was the switchboard, not the engineers.
Fixing the sales handoff means making a deal's full context travel with it instead of being recompressed into six CRM fields, so an agent can brief the next owner.
AI agents are taking over the swivel-chair layer of RevOps: CRM hygiene, billing-to-pipeline reconciliation, and forecast prep. What's left is the strategic job RevOps was promised in the first place.
AI turns coordination into ambient infrastructure, shrinking middle layers while elevating judgment, coaching, and workflow design as the new core of management.
Your next hire might not be a person. Blend humans and agents, aim freed hours at new work, and plan review as the real bottleneck.
