The Triad Is Dead. Pods Are Dead. Agent Departments Now.

Forrester and Deloitte are still publishing diagrams of "human-with-agent teams." Boxes with dotted lines connecting humans to agents inside the same pod. It looks reasonable. It's wrong.

The triad is dead. The pod is dead. What's replacing them is not pods with agents bolted on. It's a different unit entirely: a human department coordinating with an agent department on exception. Different mandate, different cadence, different career ladders.

The short version

The pod model assumed humans are the unit of execution. Agent fleets break that assumption. A pod of 5 humans plus 8 agents is structurally not a 13-person pod; the agents have different work, different cadence, different accountability. The right frame is two units: a human department doing judgment work, and an agent department doing routine and synthesis work. They coordinate on exception via a published interface.

This changes hiring (fewer humans, different roles), career ladders (less generalist execution, more specialist judgment), and operating cadence (humans on the legacy clock, agents on a continuous clock).

Where pods break

The pod breaks in three places.

Cadence first. Pods run on human meeting rhythms, standups, sprint reviews, retros. Agents work continuously. Force them into the pod's rituals and you slow them to the pod's clock. Leave them out and the pod goes blind to what they're producing. Either way the rhythm is wrong.

Then accountability. A pod assumes shared accountability across humans, and an agent can't be accountable in the human sense. So when you try to pin an agent's output on a human in the pod, you get one of two bad outcomes: one person nominally owns a hundred outputs they never produced, or nobody owns the quality of what shipped. Neither is real ownership.

And the career ladder. PM, designer, and engineer ladders were calibrated for humans doing all the work. When agents do half of it, the calibration is off. Senior PMs get evaluated on output that was never theirs. Juniors lose their path, because the routine work that used to teach the craft is now an agent department, and nobody learns the craft by reviewing it.

The replacement: a human department and an agent department

A single product squad now runs as two coordinated units.

The human department is the PM, a senior engineer, a designer, and whatever specialists rotate in. Its mandate is judgment: edge cases, customer relationships, quality standards. It keeps the weekly clock, sprint, retro, escalation review.

The agent department is the set of agents doing the squad's routine and synthesis work, the continuous monitoring, the summarization, the drafting, the classification. It runs continuously, and a human owner reads its output once a week.

Between the two sits a published interface: escalation triggers, dispute mechanisms, quality SLAs. That interface is the whole design. Get it wrong and you're back to a pod with agents bolted on.

What this changes about hiring

You hire fewer juniors, because the work that used to train them is now agent work. You hire more senior specialists, agent ops, eval engineers, prompt specialists, design system owners, none of which existed as roles in 2022. And the PM you hire in 2026 is closer to a senior strategist than a senior coordinator, because the coordination layer is the agent department's job now.

What to do this week

Pick your strongest product squad. Draw its org chart. Now draw the agents it's actually using, and ask three questions of each one: who owns its output, who reads it daily, who escalates when it's wrong.

If the answers are vague, you have a pod with agents bolted on. If they're clear, you already have a human department coordinating with an agent department, and everything above applies.

Related

• The Dual Transformation Operating Model, talent allocation and bridge engineering.
• Dual Transformation Handbook Chapter, the operating rhythm in detail.
• /cpo, the executive-strategy index.

---