Executive summary
Value comes from improving the whole operation. Define the intended outcome, connect work and decisions, then build capabilities that can be governed, measured, and improved together.
An AI initiative can meet its technical targets while leaving the business outcome unchanged. A team processes requests faster, yet exceptions still wait for another department. A useful recommendation arrives, yet nobody has authority to act on it. Local progress can coexist with an operation that remains difficult to run.
McKinsey’s August 2026 survey offers a reason to examine that gap. Nearly nine in ten respondents report regular AI use in at least one business function, while 37 percent attribute some enterprise earnings impact to AI. These are reported outcomes, not proof that any single operating choice caused the difference. [1]
The question beneath the technology is straightforward: how should this part of the business operate, and what must connect for that outcome to become dependable? The puzzle needs a shared picture before its pieces can be judged.
Each capability inherits the conditions around it. Faster analysis may depend on information that arrives late. Automated decisions may create more review work downstream. Time saved in one team may never become usable capacity if the next team remains the bottleneck.
McKinsey also reports that its AI high performers are more likely to redesign workflows fundamentally. That association supports examining how work is organized; it does not establish that redesign alone delivers value. [1]
For leadership, the practical implication is to evaluate each investment against the operation it must improve. Its contribution depends on the quality of its inputs, the decisions it can support, the receiving team’s capacity, and the cost of running and supervising it. Those connections belong in the business case.
A shared operating picture need not specify every future detail. It should make the outcome, dependencies, ownership, and constraints clear enough to guide a focused first investment.
The operating idea
Set the outcome.
Define success.
Map work and decisions.
Define connections.
Choose the right mix.
Build and test.
Measure what changes.
Improve and adapt.
Leadership owns
Outcome + priorities · Decision boundaries + ownership · Evidence for the next investment
↶ Operating evidence informs the next design decision
Whole-to-parts design begins with the result the operation must deliver, then defines how its capabilities will work together. Four moves make that discipline practical.
Define the outcome in terms an operating owner can assess: what customers or employees receive, how reliably it arrives, what it costs, and which risks must remain controlled. Describe where work begins and ends, which decisions matter, and where human judgment is required.
Set a baseline before selecting an intervention. If the goal is faster resolution, establish elapsed time, recurring delays, and rework across the full process. Task speed alone cannot show whether the operation improved.
Break the operation into connected capabilities. For each, define the inputs, outputs, required knowledge, decision authority, Outcome owner, and evidence of performance. Give equal attention to the handoffs: who accepts the result, what makes it usable, and what happens when information is incomplete or an exception appears.
This makes dependencies visible before teams commit to separate solutions. It also shows which shared foundations, such as trusted policy information or consistent records, need attention first.
Choose the combination of people, workflow, rules, conventional automation, and AI that can meet the requirement. Test technology limits, integration effort, operating cost, and review needs early. Those findings may change the design or the sequence of investment.
A focused pilot is useful when it tests a consequential part of the operating picture. Evaluate it with realistic inputs, downstream users, and exceptions so that success means more than a working demonstration.
Keep responsibilities and interfaces clear enough that a capability can be improved or replaced without leaving surrounding work unsupported. Assign ownership for monitoring performance, maintaining knowledge, resolving failures, and deciding when the design needs to change.
Expand when evidence shows that the connected operation is improving within acceptable cost and risk. A technology advance creates an option to reassess; it does not automatically justify greater autonomy.
In practice / Illustrative example
Suppose the objective is accurate, timely pay, with errors identified early and employee questions resolved clearly. A useful starting point is the flow from employee and time records through calculation, approval, payment, and correction.
In this illustrative design, AI could help classify queries, retrieve approved policy, or flag unusual inputs for investigation. Established calculation systems could continue calculating pay. Payroll specialists would retain authority over sensitive changes and unresolved exceptions.
The critical connection is what follows a flagged issue: who investigates it, which evidence they need, how a correction is approved, and whether it reaches the payment run in time. Measure payroll accuracy, unresolved exceptions, correction time, and total operating effort. A faster query response is valuable when it contributes to that wider outcome.
Executives own the picture on the puzzle box: the outcome, investment priorities, acceptable tradeoffs, and accountability for how the operation performs. Business, operations, risk, and technology leaders then work together to test whether the proposed pieces can deliver it.
Before funding or expanding an initiative, ask its sponsor to show where it fits, which dependencies must be resolved, who will own it in operation, and what evidence will justify the next commitment. Make one person accountable for the overall outcome, with explicit responsibilities across contributing teams.
This approach leaves room for experimentation. A small project can test one uncertain connection while remaining anchored to a larger operating purpose. As evidence accumulates, leaders can refine the picture and change the pieces deliberately.
For the next AI initiative you fund, can its sponsor show how it improves the complete operating outcome, which connections must work, and who will own the result?
Short, practical perspectives on AI-era operations, governance, and operating-model transformation.
Subscribe
RePerspective Labs