LeadershipNew·Falk Gottlob··updated ·4 min read

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

Forrester still talks about human-with-agent teams. Wrong unit. Agents form their own departments and interface with human departments on exception.

org designagent departmentsoperating modelcounter-canonAI native
Helpful?

FIG. 02 · DUAL TRANSFORMATION24 WEEKS · TWO CLOCKSLEGACY · 6-WEEK CYCLESPRINT 1SPRINT 2SPRINT 3SPRINT 4SUCCESSOR · 1-WEEK CYCLEBRIDGE ENG · ALTERNATING-WEEK RITUALSW00W04W08W12W16W20W24WEEKS · ONE CALENDAR · TWO CADENCES

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.

This essay is the argument for why, and what it changes about how product orgs are designed.

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).

The three breakdowns

Three sections expanding why pods break down with agent fleets:

  1. Cadence mismatch. Pods are tuned for human meeting rhythms (standups, sprint reviews, retros). Agents work continuously. Forcing agents into pod rituals slows them. Removing agents from rituals leaves the pod blind to what they're producing.

  2. Accountability mismatch. Pods assume shared accountability across humans. Agents can't be "accountable" in the human sense. Trying to assign accountability for an agent's output to a human in the pod creates either over-attribution (one human owns 100 outputs they didn't produce) or under-attribution (no one owns the quality of what shipped).

  3. Career ladder mismatch. PM/Designer/Engineer career ladders were designed for humans doing all the work. When agents do half the work, the ladder is calibrated wrong. Senior PMs are evaluated on output that wasn't theirs to begin with. Junior PMs don't have clear paths because the routine work that used to teach the craft is now an agent department.

The replacement: human department + agent department

A single product squad now has two coordinated units:

Human department: PM, Senior Engineer, Designer, plus rotating specialists. Mandate: judgment, edge cases, customer relationships, quality standards. Cadence: weekly rituals (sprint, retro, escalation review).

Agent department: the agents that handle the squad's routine and synthesis work. Mandate: continuous monitoring, summarization, drafting, classification. Cadence: continuous, with weekly review by the human owner.

The bridge between them is a published interface: escalation triggers, dispute mechanisms, quality SLAs.

What this changes about hiring

Three implications:

  1. Fewer junior hires. The work that used to train juniors is now agent work.
  2. More senior specialists. Agent ops, eval engineers, prompt specialists, design system owners. These didn't exist as roles in 2022.
  3. Different PM hiring. The PM you hire in 2026 is closer to a senior strategist than a senior coordinator. The coordination layer is the agent department's job.

What to do this week

One exercise:

Pick your strongest product squad. Draw the squad's current org chart. Now draw the agents the squad is using. Who owns each agent's 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 the answers are clear, you have a human department coordinating with an agent department, and the model in this essay applies.


Share this post

Frequently asked

What was the triad?+

The PM-Designer-Engineer unit that ran most product teams from roughly 2015-2024. Three roles, one product surface, shared accountability. Replaced by 'pods' (3-7 person cross-functional teams) at most scaleups by 2022.

Why is the pod model breaking down?+

Because pods assume humans are the unit of execution. With agent fleets shipping work, the unit shifts. A 5-person pod plus 8 agents is not a 13-person pod. It's a 5-person team coordinating with an 8-agent department. The relationship is different. The org chart hasn't caught up.

What is an agent department?+

A coherent set of agents handling related work, with a published mandate, a quality bar, and a single human owner who reads its outputs daily. Examples: Discovery (interview synthesis, NPS analysis, journey mapping), Quality (eval pipelines, drift detection, dispute classification), GTM (competitive intel, win-loss synthesis), Operations (weekly digests, retros, exec reports).

What is the relationship between human and agent departments?+

Coordination on exception, not co-execution daily. The human department owns judgment work; the agent department handles synthesis, monitoring, and routine outputs. The bridge is a published interface: when does the agent department escalate to humans, what format, what response time.

What changes about hiring?+

You hire fewer humans, and you hire them for different work. Less for 'execute the routine'; more for 'set quality standards, judge edge cases, define the agent department's mandate.' Senior product specialists, agent ops, design system owners. The IC pyramid flattens.

Keep Reading

Posts you might find interesting based on what you just read.