As AI begins to reshape organizations, many leadership teams are responding by reorganizing departments, consolidating functions, rewriting jobs, and eliminating roles based on what they believe the technology can now perform.

If the nature of work is changing, the structure surrounding it should change as well. But many of these decisions are being made from the outside in.

In his LinkedIn article, “We’re Rebuilding Org Charts Without Understanding How Roles Actually Work Anymore,” Gobekli founder Danny Done argues that organizations risk redesigning themselves around incomplete representations of how work actually happens.

Roles are more than collections of tasks

Most restructuring efforts begin with the artifacts leaders can easily see: organizational charts, job descriptions, task lists, reporting lines, and headcount.

They then ask:

• What can AI automate?
• Which activities can be removed?
• Which roles should be consolidated?
• What should the new structure look like?

Those questions are reasonable, but they miss a deeper one: How does this role actually function within a human-and-AI environment?

A role is not simply a collection of tasks. It is a pattern describing how someone thinks through problems, applies judgment, uses tools, collaborates with other people, interprets context, and contributes to decisions.

AI is changing all of those elements simultaneously.

The transformation is not limited to what gets done. It changes how work is performed, which portions are completed by people or machines, and how decisions move through a system that now includes both human and artificial intelligence.

Restructuring can simplify growing complexity

When organizations move immediately from AI capability to structural change, they risk simplifying something that has become more complex.

They may eliminate roles before understanding the human judgment, relationships, and institutional knowledge contained within them. Teams may be redesigned without visibility into how work actually travels between people and systems. New organizational charts may reflect logical assumptions about automation without representing the reality of day-to-day operations.

The primary risk is not merely inefficiency. It is misalignment.

An organization can create roles that appear appropriate on paper but do not reflect how the work functions. It can design teams that seem structurally sound but cannot operate coherently. Leaders may make workforce decisions without understanding where capability is actually located or how value is created across formal boundaries.

The org chart captures only the surface

Organizations do not yet have a consistent way to represent how roles change when AI enters a workflow, how human judgment interacts with machine-generated output, how capability appears across individuals rather than job titles, or how relationships and dependencies contribute to outcomes.

Without that visibility, leaders default to what their current systems can display:

• Structure
• Titles
• Functions
• Reporting lines
• Headcount

But these are only the visible surface of an organization.

The real structure is found in how work moves, how decisions are made, how context is shared, and how people and AI combine their capabilities to produce results.

An organizational chart can show who reports to whom. It cannot show which person provides the judgment that prevents a recurring mistake, where critical knowledge is concentrated, how one team depends upon the informal support of another, or how an AI system is changing the division of responsibility within a role.

Those relationships may be invisible, but they are often where organizational value resides.

Understand the work before redesigning the structure

The answer is not to resist AI or slow organizational change unnecessarily. It is to become more precise about what is being changed.

Before asking what the new structure should look like, leaders need to understand what the work is becoming. They need ways to observe how roles are evolving, how human and machine capabilities interact, where judgment remains essential, and how value moves across the organization.

Only then can organizational design reflect the system people and AI are actually creating together.

Rebuilding the org chart may still be necessary. But without a deeper understanding of how roles and relationships function, it will remain a partial solution—one that reorganizes the visible structure without necessarily improving the work underneath it.

Read more on Linkedin