What a role owns is not the same as what a role is there to advance. That gap is where structural coherence begins to break down, not dramatically but in the accumulating cost of slower decisions, more escalation, and growing dependence on a handful of capable people to produce the alignment the structure itself should be generating.
Most role design stops at ownership — who is responsible, who is accountable, who gets consulted and informed. That answers the wrong question, and you can watch it fail in real time.
Responsibility tells you what you own. Advocacy tells you what you owe.
I once worked with an online university where five roles all believed they owned the same decision, and three separate RACI exercises failed to settle it. The RACIs were not the problem. Nobody had ever defined what each role was supposed to advance on behalf of the enterprise — so capable people kept defending territory the structure had never assigned them.
A concrete pattern
A large online university in the middle of a multi-year agile transformation found itself with five roles in open conflict: Project Manager, Product Owner, Product Manager, Scrum Master, and Program Manager. Each felt the others were encroaching. Prioritization, stakeholder management, dependency resolution, delivery quality: all contested.
Leadership tried to resolve it three separate times. Three different RACIs, produced by a director, a manager, and a vice president. None of them worked.
The roles weren't unclear. Every role had a title, a description, and an owner. What the organization had never answered, for any of the five roles, was why it needed that role to exist. Not the work. The purpose. Once that question was answered, the conflict dissolved. Not because the roles changed. Because the people in them finally understood what each role was there to stand for, and why.
This pattern is not specific to higher education, and it is not specific to agile. I see it across regulated industries, scaling product organizations, and large enterprises partway through any structural change. The symptoms rhyme. The cause is the same.
The distinction most role design misses
The instruments used to clarify roles, the RACIs and org charts and job descriptions, are useful, but they describe the wrong layer. They describe what work gets done and who is on the hook for it. They do not describe what the role is there to stand for when the work runs into friction.
The cost of that gap rarely shows up as a single failure. It shows up as drift: slower decisions, more escalation, and growing dependence on a handful of capable people to produce the alignment the structure itself should be generating.
What role advocacy is
Every meaningful role carries a sphere of advocacy: a decision domain, a set of standards, a flow of work, a customer outcome, or the integrity of a function that the enterprise as a whole depends on.
The person in that role isn't merely assigned to that sphere. They are positioned to advance it.
Responsibility tells you what you own. Advocacy tells you what you owe.
Roles are designed as clear spheres of advocacy derived from enterprise purpose. People know what their role exists to preserve, advance, or challenge on behalf of the broader system, and they hold unambiguous decision authority within that scope.
Spheres of advocacy are derived from organizational purpose, not invented role by role. A role's sphere exists because the enterprise needs certain conditions to be created and sustained, and has positioned a role to carry that work. Without clarity on purpose, role design defaults to habit, precedent, or job market convention, producing roles that describe a profession rather than a position in this system, serving this enterprise, advancing this purpose.
Role advocacy is not a synonym for accountability. Accountability defines who owns a result. Advocacy defines what a role is there to actively preserve, advance, and challenge on behalf of the system. An organization can have perfectly clear accountability and still have no one advocating for the conditions a role exists to preserve. Both matter. They are not the same.
The pattern when advocacy is missing
When accountability is as far as a role extends, a specific pathology follows. People define their work defensively. Local targets get protected at the expense of system outcomes. Escalation becomes the default when problems cross role boundaries, not because people are seeking guidance but because they need someone with authority to absorb the risk.
The Organizational Entropy & Extropy Model names this as one of the core entropy patterns: accountability without advocacy produces blame avoidance, territorial behavior, and escalation as a means of validation.
The Heroic Organization
The most common expression of this failure is what the Extropy Model calls the Heroic Organization.
When no one is explicitly positioned to advocate for certain outcomes, those gaps do not stay empty. Capable, committed people fill them. They translate across functions that should naturally be aligned. They make decisions unilaterally, not because authority has been clearly granted but because the organization has built a silent dependency on their judgment. When that judgment meets resistance, heroes escalate, often not to seek clarity but to secure legitimacy. Informal authority accumulates in individuals the structure was never designed to contain.
The hero is not a reluctant participant. They have been reinforced: praised, rewarded, validated at senior level, repeatedly. From where they stand, that looks like trust. It is also a trap.
When the heroic performance finally breaks down, the organization does not examine the condition that required heroics. It looks for someone to hold accountable.
The hero, who believed they were positioned to be admired, discovers they were also positioned to absorb blame. Attention shifts from the structural question (why did we need this person to carry so much?) to the personal accusation: I thought we could rely on them. I guess not.
Role advocacy in the age of agents
This is where the conversation about AI and agentic workflows enters a different register.
Most AI adoption so far has been framed as task automation. An agent takes a workflow, executes it, and produces outputs. Someone is still in the role. The agent is a tool.
But agentic systems are increasingly operating in a role. They make decisions within a boundary. They interact with other roles. They produce work that directly advances (or fails to advance) an organizational outcome. At that point, the agent is no longer a tool inside a role. It is occupying one.
Here the distinction between responsibility and advocacy becomes design-critical.
When a human occupies a role, they bring implicit advocacy with them. A capable person in an ambiguously defined role will often invent the advocacy the organization failed to design. Sometimes for better. Often, as the Heroic Organization pattern shows, for worse.
An agent cannot do that. An agent will carry responsibilities precisely and advocate for nothing.
Handing an agent a role's tasks without handing it the role's advocacy scales the exact pattern this essay has been describing. Work gets done. Decisions stall. The organization becomes dependent on the humans adjacent to the agent to supply the advocacy the agent's role is silently missing.
The implication for role design is sharp. Handing responsibilities to an agent forces a question most organizations have never explicitly answered about their human roles: what is this role here to advance, beyond what it produces? That question can no longer be deferred to the judgment of whoever happens to be in the seat.
Designing roles for agents (or redesigning human roles around agentic collaboration) requires making advocacy explicit. For an agent, that means encoding what the role is there to advance, defining the constraints and evaluations that let it challenge inputs that threaten that advocacy, and establishing the standing it has to interact with other roles as more than a task executor. For a human working alongside agents, it means understanding which elements of their role's advocacy have been absorbed, amplified, or left uncovered, and redesigning accordingly.
There is also an emerging version of the Heroic Organization in this territory. A single capable agent gets used across many domains. The organization builds silent dependency on its outputs. When the agent eventually fails, or gets something subtly wrong at scale, attention turns to the agent ("AI isn't ready") rather than the structural condition that required it to carry advocacy no one had defined.
Role advocacy in practice
When humans and agents operate within roles understood as advocacy positions, something changes. Decision-making improves because the basis for judgment is clearer. Escalation drops because those operating within their advocacy have both the standing and the obligation to resolve tension at the level it actually exists. Cross-functional work improves because roles are active, not passive.
A chart can assign a role. A job description can describe it. A prompt can instruct an agent. None of these substitutes for an occupant, human or agent, who understands what the role exists to stand for, and acts accordingly when the system needs it.
Role advocacy is one of the sub-themes of Structural Coherence in the Organizational Entropy & Extropy Model, which assesses organizations on two dimensions. Structural Coherence asks whether the structure reliably generates alignment without constant management intervention. Adaptive Capability asks whether the organization has the developed ability to learn, adapt, and grow as a system, not just the people inside it. Advocacy gaps are a coherence failure. They also suppress Adaptive Capability, because an organization cannot develop what no one is positioned to steward.
Designing roles as advocacy positions begins with a question most organizations have never formally asked: what is each role positioned to preserve, advance, and challenge on behalf of the enterprise? Answering it requires clarity on organizational purpose first, then an honest look at where the current structure falls short: where roles are compensating for each other, where escalation is functioning as structural glue, and where heroic individuals, or heroic agents, are carrying work the organization's design should be generating.
The diagnostic questions
The questions are concrete. A leadership team can sit with them for thirty minutes and surface more than enough to act on.
- Where is escalation functioning as structural glue rather than genuine conflict resolution?
- Which roles, human or agent, are currently compensating for advocacy gaps that should never have existed?
- Which AI outputs sit in the organization without a role positioned to champion them?
- Which of your most capable people are carrying advocacy the structure never defined, and what happens when they leave?