Rethinking the Organization | Article 3 of 4

Stop Automating the Process. Understand the Intent.

Enterprise software taught organizations to define the path through work. AI can make parts of that path more adaptive, but only when the organization is clearer about the outcome, decisions, constraints and accountability the process exists to protect.

The workflow is visible. The intent is often not.

Consider the creation of a new position. In a mature Workday environment, the request may move from the manager to HR, Finance, compensation and recruiting. Information is validated, funding is confirmed, organizational design is reviewed and authority is recorded before the role can be opened.

An intelligent capability can assemble the business context, check that required information is present, prepare materials and coordinate the route. Those improvements are useful. But if the capability simply accelerates the existing sequence, it may also reproduce duplicated review, obsolete controls and participation that exists mainly to move information.

The more consequential design move is to identify what the process is meant to accomplish and protect. The visible steps are only one expression of that intent. They are not the intent itself.

Enterprise platforms created enormous value by making work explicit. Workday business processes encode initiation, conditions, security, reviews, approvals, evidence, exceptions and completion. That architecture made complex work more repeatable, controlled and auditable, while exposing differences between policy, practice and assumed authority.

Over time, however, organizations became highly fluent in the route through the work. A team can often recite the sequence without being equally clear about the decision each step makes, the risk it controls or the contribution each participant provides.

A process is a mechanism for producing an outcome. It should not become a substitute for understanding the outcome.

Process can outlive its purpose

A position request with three approvals may contain three legitimate decisions. One establishes that the role is needed, another confirms that funding exists and a third protects the integrity of the organizational design. Each decision may require different evidence, expertise and authority.

The same process can also conceal a less coherent design. All three approvers may review essentially the same information. One step may have been added during a temporary cost-control initiative. Another may compensate for a limitation in a former system. The route remains after the original reason has disappeared.

AI can perform an approval, prepare a recommendation or move the transaction more quickly. Speed alone does not distinguish a necessary control from a historical habit. The stronger design separates the decisions, identifies the evidence each one needs, assigns clear ownership and makes the consequence being managed visible.

A process should be the consequence of clear decisions, not the place where unclear decisions go to hide.

AI makes intent more executable

Traditional software largely required a prescribed route. The organization defined the conditions in advance and the system followed them. AI can increasingly work from a clearer statement of purpose: an objective, relevant context, authoritative sources, permitted tools, decision boundaries and defined points of human intervention.

This is not permission to give an agent a vague instruction and let it invent the operating model. Useful intent is more precise than a long workflow. It establishes the outcome, the trade-offs that matter, the policies and controls that apply, the actions the system may take, the situations it must escalate and the conditions under which it should stop.

The path can become more flexible only because the architecture around it becomes clearer. AI may reduce the need to prescribe every mechanical step, but it increases the need to define what must remain true regardless of the route taken.

Adaptive execution requires more clarity about purpose and boundaries, not less.

Decision architecture comes before workflow

A useful process redesign starts one level above the current sequence. Four elements make the underlying intent visible and give both people and intelligent systems a stable basis for execution.

The workflow should emerge from these choices rather than substitute for them.

Hiring is a collection of decisions

Hiring is often presented as one journey from requisition through sourcing, screening, interviewing, selection and offer. In practice, it contains several different decisions: the position should exist, funding is available, the role is designed appropriately, the candidate meets the relevant standard, the selection is justified and the proposed terms are authorized.

AI can coordinate much of the activity around those decisions. It can assemble the approved position and workforce context, validate required evidence, schedule participants, prepare interview materials, identify a case outside normal parameters and draft an offer within established boundaries.

The organization still defines the decision architecture. Workday can provide identity, position, organization, compensation, security and business-process context. HR, Finance and business leaders still establish the standards, decision rights, policies and accountability that the capability must respect.

The route may become shorter and vary by case. The intent should remain stable enough that the organization can explain why the work took that path and who owned the result.

When the path becomes more flexible, the architecture of decisions has to become more explicit.

The future combines prescribed and adaptive work

AI is not the end of workflow. Some processes should disappear, some should become shorter and many universal approvals can move toward exception-based involvement. Other work should remain deliberately prescribed because the sequence itself protects evidence, separation of duties, fairness or control.

Most enterprise work will combine all three modes. Decision rights and control boundaries remain fixed, routine activity adapts to the context of the case and human judgment enters when uncertainty or consequence makes it valuable.

Participation should reflect contribution

Mature processes often include approvers, reviewers, observers, notification recipients and people who receive a report after the work is complete. Some contribute evidence or expertise. Some own a decision or a material control. Others are present because the process historically served as the only reliable way to move information or preserve visibility.

AI makes those contributions easier to separate. A person who only reformats information, passes it onward or confirms a fact the system can already establish may no longer need to sit in the path. A person who contributes judgment, relationship, authority or accountability should remain, with that contribution made more explicit than a generic approval label allows.

The objective is not the removal of people. It is the deliberate placement of human attention where it materially improves the outcome.

Human participation should be designed around contribution, not inherited from the route.

Enterprise platforms become context, not merely destinations

As AI participates across applications, people may spend less time navigating the interface of any single system. That does not reduce the importance of Workday. Worker and position records, supervisory and financial structures, roles, security, business processes and effective-dated changes provide the organizational context that determines who may see information, which decisions a role can make and where an action belongs.

Current configuration should still be treated as evidence, not proof. It records how the organization operates today; it does not establish that every approval, exception or report should survive. Workday can anchor identity, authority and organizational truth while the mechanics of execution are redesigned.

What adaptive process design requires

Design the work around what matters

Automating an inherited process can improve speed without improving the work. Redesign begins when the organization separates the outcome from the route, distinguishes the decisions hidden inside the approval chain and makes the necessary controls and human contributions explicit.

The process does not disappear. It becomes the expression of intent rather than its substitute. AI may reduce the effort required to prescribe every step, but it raises the level at which architecture must be designed: purpose, authority, context, evidence and responsibility.

The final article considers the organizational consequence: roles and management structures built around moving information become open to redesign as intelligence interprets context, coordinates routine work and directs attention to exceptions.