AI is changing who does the work, how decisions are made, and what managers must contribute. But technology cannot decide what the enterprise should value, which trade-offs it should accept, or who remains accountable when the decision produces consequences.
Tuesday, October 27, 2026 | Leadership & Organizational Value Creation
Yesterday we argued that organizations should stop treating the 2027 budget as the operating system.
The budget allocates money.
The operating system must connect:
- Strategy.
- Assumptions.
- Capital.
- Ownership.
- Execution.
- Exceptions.
- And learning.
But once that architecture exists, another question immediately appears:
Who actually makes the decisions inside it?
That question is becoming more complicated.
Artificial intelligence can now analyze information, generate alternatives, forecast outcomes, recommend actions, prioritize work, identify exceptions, draft plans, coordinate tasks, and increasingly take action without waiting for a human at every step.
At the same time, organizations are questioning how many management layers they need, what middle managers should actually do, which decisions should remain centralized, which decisions should move closer to the work, and whether AI allows companies to operate with fewer people.
Those are important questions.
But there is a danger in answering them primarily through a technology lens.
The leadership challenge is not simply deciding what AI can do.
It is deciding what humans must continue to own.
AI Is Already Inside the Decision Process
Deloitte’s 2026 Global Human Capital Trends research found that 60% of executives regularly use AI to support decisions.
Yet only 5% of organizations consider themselves leaders in AI-enabled decision-making.
Deloitte’s related decision-intelligence research also found that 57% of organizations operate at low levels of decision-making maturity.
That combination should concern leadership teams.
AI adoption is moving quickly.
Decision disciplines are not necessarily moving with it.
The technology is becoming more capable at exactly the moment many organizations still struggle with unclear decision rights, too many approvals, weak accountability, poor escalation, slow cross-functional coordination, low-quality data, unclear risk thresholds, and inconsistent management judgment.
AI does not automatically repair those weaknesses.
In some cases, it amplifies them.
Deloitte makes the point directly: as AI expands into organizational decision-making, companies risk opaque choices, diluted accountability, and erosion of human agency unless they intentionally design how humans and machines interact as decision makers.
That should change how leaders approach AI.
Do not automate a decision before you understand the decision.
Start With the Decision, Not the Technology
Many AI implementations begin with a question like:
“Where can we use AI?”
That is probably the wrong starting point.
Begin instead with:
What decision are we trying to improve?
Then ask:
- Who owns it today?
- What information does that person need?
- How frequently is the decision made?
- What is the cost of delay?
- What is the cost of a wrong decision?
- What judgment does it require?
- What rules can be standardized?
- What uncertainty remains?
- What happens if the decision fails?
Only after those questions should leadership decide how much authority AI receives.
Build a Human–AI Decision Architecture
Not every decision requires the same relationship between people and AI.
A useful 2027 operating model should distinguish at least four decision categories.
1. AI Advises. Human Decides.
This is appropriate when the decision contains substantial ambiguity, strategic consequence, ethical consideration, customer impact, talent implications, reputational risk, or irreversible capital commitment.
AI might analyze market scenarios, acquisition targets, supplier risks, capacity alternatives, customer profitability, or organizational options.
But the human leader retains explicit authority.
AI improves the evidence. Leadership owns the choice.
2. AI Decides Within Defined Guardrails
Some decisions can increasingly be automated.
Examples might include routine replenishment, scheduling adjustments, low-risk pricing recommendations, workflow routing, fraud screening, maintenance prioritization, or inventory repositioning.
But automation requires clearly defined boundaries.
- What can AI decide?
- Within what financial limit?
- Using which data?
- At what confidence threshold?
- Under which conditions must it stop?
- When must it escalate?
- Who monitors its performance?
Delegated authority does not eliminate accountability.
The human owner remains responsible for the system.
3. AI Detects the Exception. Manager Exercises Judgment.
This may become one of the most powerful models.
AI continuously monitors the system.
Most activity stays within normal limits.
Then something changes.
Demand deviates. A supplier deteriorates. Margin moves. A customer behaves differently. Capacity tightens. Inventory rises. A cyber pattern appears.
AI identifies the exception.
The manager determines what it means.
AI detects abnormality. Humans interpret consequence.
This model may allow organizations to move from supervising routine work toward managing exceptions.
That could profoundly change the manager role.
4. Human Override Is Always Possible—and Always Visible.
Every significant automated decision system needs an override mechanism.
But overrides should not disappear.
If a leader overrides the AI recommendation, record who made the decision, why, what information changed the judgment, and what happened afterward.
That information becomes learning data.
Over time, the organization can discover when humans consistently outperform the model, when the model consistently outperforms humans, which assumptions repeatedly fail, and where decision rules need redesign.
The override should become part of the learning system—not an invisible exception.
Every Decision Needs an Accountability Statement
For every material AI-enabled decision, complete this sentence:
“AI may ________, but ________ owns the outcome.”
For example:
- AI may recommend supplier allocation, but the Supply Chain VP owns continuity and service.
- AI may optimize production scheduling, but the Plant Manager owns output and customer commitments.
- AI may identify investment scenarios, but the CFO owns the capital recommendation.
- AI may score acquisition targets, but the CEO and board own the transaction decision.
This sounds obvious.
But clarity becomes more important as automation increases.
The more invisible the decision process becomes, the more explicit accountability must become.
Decision Speed Is Not Decision Quality
One of AI’s greatest advantages is speed.
That is also one of its risks.
An AI system can process information at a scale and velocity far beyond human capability.
But organizations should not confuse:
Faster decisions
with:
Better decisions.
A decision can be fast and wrong. Fast and poorly governed. Fast and impossible to explain. Fast and strategically inconsistent. Fast and locally optimal but harmful to the enterprise.
That is why the objective should not be:
Automate as many decisions as possible.
It should be:
Improve the speed, quality, consistency, and accountability of decisions that create enterprise value.
Those are different objectives.
The Middle Manager Is About to Become More Important—not Less
Much of the conversation around AI and organizations has focused on whether technology will eliminate management layers.
That question is understandable.
But it may be premature.
McKinsey’s October 2026 research argues that agentic AI is changing three constraints that shaped traditional organizations:
- Human capacity.
- Expertise.
- Coordination.
That opens the door to redesigning jobs, teams, spans of control, management layers, and hierarchies.
But McKinsey also highlights a significant gap between personal productivity and enterprise results.
- 80% of respondents say AI has improved individual productivity.
- Only 37% report positive EBIT impact.
- And only 6% qualify as AI high performers.
That gap tells us something.
Giving people faster tools does not automatically create a faster enterprise.
Someone must redesign the work around the technology.
And much of that responsibility will fall to managers.
HBR: AI Is Increasing the Pressure on the Middle Layer
Harvard Business Review researchers examining AI adoption found a recurring pattern.
Senior leaders see strategic opportunity.
Junior employees experience significant productivity gains.
But middle managers inherit additional work.
They must validate AI outputs, identify errors, maintain quality, coach employees, teach new workflows, interpret leadership expectations, redesign processes, manage resistance, and still deliver the original operating results.
Often without additional support.
The HBR research describes middle managers as increasingly overloaded by AI adoption rather than simply replaced by it.
That should force a leadership rethink.
The middle manager may not be the layer AI eliminates.
It may be the layer that determines whether AI produces enterprise value.
Stop Asking Managers to Do the Old Job Plus AI
This is one of the most common transformation mistakes.
Leadership introduces a major new technology.
Then tells managers:
- Keep the same responsibilities.
- Hit the same targets.
- Attend the same meetings.
- Produce the same reports.
- Manage the same number of people.
- And now also learn AI, train the team, validate outputs, redesign workflows, govern risk, and find productivity.
That is not transformation.
That is workload accumulation.
If AI changes the work, management roles should change too.
Redesign the Manager Role for 2027
The traditional manager was often expected to distribute tasks, collect information, approve work, track status, review output, escalate exceptions, control access to information, and report results upward.
Much of that work can increasingly be assisted or automated.
The 2027 manager should move toward higher-value responsibilities.
From Task Allocator to Workflow Architect
Instead of deciding who performs every task, managers should increasingly decide which work should be automated, which work requires human judgment, where AI should assist, where humans should review, and what sequence produces the best outcome.
Managers increasingly design the system in which work happens.
From Information Collector to Judgment Provider
Managers should spend less time collecting data that systems can generate automatically.
They should spend more time asking:
- What does this signal mean?
- Which risk matters?
- What assumption changed?
- What trade-off should we make?
- What deserves escalation?
From Approval Layer to Exception Manager
If every routine decision still requires management approval after AI implementation, the organization may have automated analysis without redesigning governance.
Routine work should increasingly flow within defined guardrails.
Managers should focus on the cases where conditions change, rules conflict, risk increases, or judgment matters.
From Status Reporter to Coach
AI can increasingly generate the status report.
The manager should use the time saved to develop judgment, improve problem solving, give feedback, build capability, and prepare people for increasingly complex work.
From Knowledge Gatekeeper to Human–AI Orchestrator
Historically, managers often gained value because they possessed information others did not.
AI makes knowledge dramatically more accessible.
Managerial value should increasingly come from context, judgment, coordination, coaching, interpretation, and creating alignment around action.
From Activity Manager to Outcome Owner
This may be the most important shift.
Managers should be evaluated less on how much activity they supervise and more on what results their operating system consistently produces.
AI should reduce management administration—not management accountability.
There Is Another Risk: Who Develops the Next Generation of Leaders?
This deserves more attention than it receives.
Many early-career tasks are repetitive.
Those are exactly the tasks AI can automate fastest.
But those tasks have historically also provided context, pattern recognition, customer exposure, technical understanding, business judgment, and organizational learning.
A young analyst may have learned by preparing the report. A junior engineer by analyzing failures. A planner by manually balancing demand and capacity. A consultant by constructing the first draft. A buyer by comparing suppliers.
If AI performs the entire task, the organization gains productivity.
But what happens to the learning?
Efficiency can accidentally remove apprenticeship.
Managers must deliberately replace developmental experiences that automation removes.
That might mean decision simulations, rotations, shadowing, scenario exercises, structured reviews, human override responsibilities, postmortems, and deliberate exposure to increasingly complex decisions.
The future manager must develop people for judgment—not merely train them to use tools.
The CEO Still Cannot Delegate the Hardest Choices
AI may support decision-making.
Managers may redesign work.
Leadership teams may develop strategic options.
But one responsibility remains fundamentally difficult to delegate:
Strategic choice.
McKinsey’s October 2026 research argues that strategy is a team sport but identifies six strategic responsibilities CEOs should not delegate.
They include choosing how personally to engage in strategy, owning the value-creation thesis, making the hardest resource-reallocation decisions, changing strategic tempo when necessary, helping the organization see opportunity in uncertainty, and acting as a catalyst for change.
This matters directly to the 2027 planning process.
The CEO Can Delegate Analysis. Not the Choice.
A leadership team may analyze markets, competitors, customers, technology, capital requirements, risks, acquisitions, capacity, and scenarios.
AI may make that analysis faster.
But eventually someone has to choose.
Should we build the plant? Buy the competitor? Exit the product? Enter Mexico? Increase inventory? Reduce management layers? Invest in AI? Stop the ERP project? Close the facility? Change the leadership team? Protect margin or protect market share? Accept the risk or fund resilience?
Those decisions cannot be resolved simply by averaging executive opinions.
Strategic leadership requires choice.
McKinsey: Structured CEO Strategy Practices Matter
The McKinsey research cites a 2026 Management Science study finding that companies led by CEOs using more structured approaches to strategy experienced stronger performance.
A one-standard-deviation increase in the structure of CEO strategic practices was associated with approximately 9.5% higher five-year growth, with particularly strong effects in industries facing greater strategic complexity.
That does not mean the CEO should dominate every strategy discussion.
It means the CEO cannot become a spectator.
A strong strategic process should produce diverse perspectives, rigorous analysis, constructive challenge, real alternatives, and then:
A decision.
Consensus Is Not Strategy
One danger in large organizations is that strategy becomes an aggregation exercise.
Sales wants growth. Operations wants capacity. IT wants digital investment. HR wants capability investment. Finance wants efficiency. Supply Chain wants resilience.
Each priority is reasonable.
Leadership combines them.
Then everything becomes strategic.
That is not strategy.
Strategy is what survives the trade-off.
The CEO must eventually ask:
- Which three outcomes matter most?
- Which initiative loses funding?
- Which market will we not enter?
- Which customer segment will we not prioritize?
- Which capability deserves disproportionate investment?
- Which risk are we willing to accept?
- Which assumption are we willing to challenge?
- What are we prepared to stop doing?
Those are uncomfortable questions.
That is why they cannot be delegated indefinitely.
The CEO’s Most Difficult Job May Be Reallocation
Organizations become emotionally attached to resources.
Budgets. People. Plants. Products. Programs. Technologies. Customers. Historical businesses.
The CEO has a unique responsibility to challenge that attachment.
Yesterday’s Monday Execution post argued that 2027 planning should include active capital reallocation.
Tuesday adds the leadership dimension.
Capital does not move simply because Finance identifies a better return.
Someone must overcome organizational politics, historical commitments, executive ownership, sunk-cost thinking, and fear of admitting that a previous decision is no longer optimal.
That is leadership work.
Build a Three-Level Decision System
The combined lesson from AI, managers, and CEO strategy is that organizations need different decision levels.
Not everything should reach the CEO.
Not everything should be delegated to AI.
And not everything belongs with middle management.
A strong 2027 operating system might establish three levels.
Level 1 — Automated Operating Decisions
These decisions are frequent, low risk, highly rule based, reversible, and measurable.
AI can increasingly manage them within clear guardrails.
Examples: routine scheduling, workflow routing, inventory replenishment, low-value transactions, and standard service responses.
But every automated decision still needs a system owner, performance monitoring, an escalation trigger, and override authority.
Level 2 — Managerial Judgment Decisions
These decisions involve exceptions, trade-offs, ambiguity, cross-functional coordination, people, customers, or operational risk.
Managers should increasingly own these.
Examples: production recovery, supplier exceptions, customer prioritization, capacity conflicts, workforce allocation, quality deviations, and AI overrides.
Managers should receive enough authority to resolve them quickly.
If every exception still travels three levels upward, the organization has not redesigned decision rights.
Level 3 — Enterprise Strategic Choices
These decisions materially change capital, portfolio, risk, business model, competitive position, leadership structure, or enterprise direction.
Examples: acquisitions, major CapEx, market entry, footprint, portfolio exits, large transformation programs, strategic partnerships, and major restructuring.
These require active CEO involvement and often board oversight.
Technology can improve these decisions. It should not obscure who owns them.
Decision Rights Should Follow Consequence
A useful rule:
The greater the consequence, the more explicit the human accountability.
For every decision ask:
- How reversible is it?
- What financial value is at risk?
- What customer consequence exists?
- Could people be materially affected?
- Is the decision explainable?
- Does it create regulatory exposure?
- Does it change strategy?
- What happens if AI is wrong?
Then determine the required level of human authority.
Create Trust Thresholds
Deloitte’s 2026 Human Capital Trends emphasizes intentionally designing human–AI relationships rather than allowing them to evolve informally.
One practical approach is to establish trust thresholds.
- Low consequence + high AI confidence — Automate.
- Moderate consequence + high AI confidence — AI acts; human monitors.
- High consequence + high AI confidence — AI recommends; human approves.
- Any consequence + low AI confidence — Escalate.
- Material ethical, regulatory, talent, or strategic consequence — Explicit human decision required.
This provides a much better governance model than simply asking:
“Can AI do this?”
Build an AI Decision Ledger
For your most important AI-enabled decisions, maintain a simple record.
- Decision — What is being decided?
- Frequency — How often?
- AI Role — Recommend, decide, monitor, or escalate?
- Human Owner — Who owns the result?
- Guardrail — What limit cannot be exceeded?
- Escalation Trigger — When must the decision move to a human?
- Override Authority — Who can intervene?
- Outcome Metric — How do we know the decision was good?
- Learning — What does the system change when outcomes differ from expectations?
That turns AI governance into management practice.
The Leadership Team Must Also Decide What AI Should Not Decide
Technology enthusiasm can create pressure to maximize automation.
That is not necessarily wise.
Some decisions carry value precisely because they require human judgment.
Consider promotions, terminations, leadership succession, strategic acquisitions, ethical conflicts, customer relationship decisions, major restructuring, plant closure, entering a politically sensitive market, and changing the company’s risk appetite.
AI can inform those decisions.
But leadership should think carefully before delegating human consequence to an algorithm.
Efficiency is not the only value an organization should optimize.
Redesign Meetings Around Decisions
If decision-making is becoming a formal organizational capability, meetings should change too.
Many executive meetings are organized around presentations.
Business review. Sales review. Operations review. Technology review. Finance review.
Each function presents information.
But what decision needs to be made?
A stronger meeting agenda starts with:
Decision required
Then:
- Evidence.
- Alternatives.
- AI recommendation if applicable.
- Risks.
- Owner.
- Decision.
- Next action.
That can dramatically improve leadership time.
Separate Information From Decision
If the executive team receives a 70-page deck for a decision that could be expressed in one sentence, the organization may have confused information volume with decision quality.
Try this format:
- Decision — What specifically must leadership decide?
- Recommendation — What is being proposed?
- Alternatives — What else could we do?
- Economics — What value is created?
- Assumptions — What must be true?
- Risk — What could go wrong?
- AI Contribution — What analysis or recommendation came from AI?
- Human Judgment — What remains uncertain?
- Owner — Who executes?
- Review Trigger — What would make us reconsider?
That is decision architecture.
The Manager’s Role in the Decision System
Managers should not simply transmit decisions from executives downward.
They should become a critical judgment layer.
A strong manager should know which decisions they fully own, which decisions AI can execute, which decisions require cross-functional consultation, which decisions must be escalated, and which risks cannot be accepted locally.
That clarity can reduce meetings, emails, escalations, approval delays, and executive bottlenecks.
Leadership scale comes partly from pushing the right decisions to the right level.
The CEO’s Role in the Decision System
The CEO should not become the final approver of every difficult issue.
That creates dependency.
The CEO should instead make three things clear.
- What only the CEO can decide.
- What the executive team should decide.
- What should never need CEO involvement.
That creates organizational speed.
The Board’s Role
Boards should ask a different set of questions as AI expands.
Not simply:
“Are we using AI?”
Instead:
- Which material decisions involve AI?
- What authority has been delegated?
- What decisions require human approval?
- How is model performance measured?
- Who owns failure?
- How are overrides recorded?
- What happens when the system is uncertain?
- Where might AI create regulatory, ethical, reputational, or talent risk?
Board oversight should focus on consequences, not technology enthusiasm.
The 2027 Leadership Architecture
Put the three themes together.
The future operating model needs three capabilities.
1. AI That Expands Decision Capacity
Use AI to analyze, detect, recommend, automate appropriate work, and accelerate response.
2. Managers Who Convert Capacity Into Results
Managers must design workflows, exercise judgment, coach people, manage exceptions, coordinate across functions, and protect accountability.
3. CEOs Who Make the Choices That Define the Enterprise
CEOs must choose priorities, allocate capital, accept or reject risk, make portfolio choices, set strategic tempo, and decide what the organization will stop doing.
Together:
AI creates capacity.
Managers create coordination and judgment.
CEOs create strategic direction and trade-offs.
None replaces the others.
A 30-Day Leadership Reset for AI-Enabled Decisions
Week 1 — Map the Decisions
Identify the 20 decisions with the greatest impact on revenue, margin, cash, customer service, risk, talent, and strategic execution.
Do not begin with technology.
Begin with decisions.
Week 2 — Define Decision Rights
For each decision identify AI authority, manager authority, executive authority, escalation trigger, override rights, and accountable owner.
If ownership is unclear, fix it before automating.
Week 3 — Redesign the Manager Role
Identify management activities that AI can reduce or eliminate.
Then deliberately replace them with coaching, exception management, workflow design, cross-functional coordination, and judgment development.
Do not simply add AI to the existing job.
Week 4 — Clarify CEO-Level Choices
Identify the five decisions in 2027 that leadership must not allow the organization to dilute into consensus.
For each: define the trade-off, quantify the value, challenge the assumptions, create alternatives, set the decision date, and make the choice.
Questions for Your Executive Team
- Which important decisions are currently unclear about who owns the consequence?
- Where are we automating activity without redesigning decision rights?
- Which decisions could AI safely make within predefined guardrails?
- Which decisions should always require explicit human judgment?
- What happens when a manager disagrees with the AI recommendation?
- Are overrides captured and used as organizational learning?
- Which middle managers are carrying additional AI responsibilities without having old responsibilities removed?
- What management work should disappear because AI now performs it better?
- How will junior employees develop judgment if AI performs the work where they historically learned it?
- Which 2027 strategic choice is the leadership team avoiding because there is no comfortable consensus?
- Which resource should be reallocated even though an executive currently owns it?
- What decision can only the CEO make—and when will it be made?
These are not technology questions.
They are leadership questions.
The ENLOSA Perspective
At ENLOSA: Global Business Solutions, we call the connection between strategy and execution Stratactic™.
AI changes the mechanics of that connection.
It makes information faster.
Analysis faster.
Coordination faster.
And increasingly action faster.
But leadership still determines what matters, what risk is acceptable, where capital goes, who owns the outcome, when the organization changes direction, and what the enterprise refuses to delegate.
AI can accelerate the operating system.
It cannot supply its purpose.
That remains leadership’s job.
The strongest 2027 organizations will not simply deploy the most AI.
They will build the clearest architecture between machine intelligence, human judgment, managerial authority, executive accountability, and strategic choice.
That is where AI begins to create enterprise value instead of simply individual productivity.
Tuesday Executive Call to Action
Take one important 2027 decision this week.
Not a routine transaction.
A decision that materially affects capital, customers, people, risk, or strategy.
Put it on one page.
- Decision — What exactly must be decided?
- AI Role — What may AI analyze, recommend, or decide?
- Manager Role — What judgment belongs closest to the work?
- Executive Role — What requires senior authority?
- Owner — Who owns the final outcome?
- Guardrails — What cannot be exceeded?
- Escalation Trigger — What requires human intervention?
- Override — Who can change the recommendation?
- Learning — How will the outcome improve the next decision?
Then ask:
If this decision goes wrong six months from now, will everyone know who owned it—or will the organization blame the algorithm, the process, or each other?
If the answer is unclear, the leadership architecture is incomplete.
Fix that before scaling the technology.
AI can make the enterprise faster.
Clear leadership is what keeps faster from becoming directionless.
References
- Deloitte — 2026 Global Human Capital Trends: AI and the Future of Human Decision-Making. Deloitte reports that 60% of executives regularly use AI to support decisions, while only 5% of organizations consider themselves leaders in AI-enabled decision-making. Its decision-intelligence research found 57% of organizations operating at low decision-making maturity. Source
- Deloitte — Getting Human and Machine Relationships Right. Deloitte argues that organizations need to deliberately design human–AI relationships, including decision authority, accountability, trust, and human agency. Source
- McKinsey & Company — “AI Is Changing Work. Now It Has to Change the Organization,” October 5, 2026. McKinsey reports that AI is changing organizational constraints around capacity, expertise, and coordination. While 80% of respondents report higher individual productivity, only 37% report positive EBIT impact and 6% qualify as AI high performers. Source
- Harvard Business Review — “AI Adoption Is Overloading Your Middle Managers,” June 26, 2026. Research found middle managers absorbing additional responsibilities for AI validation, error identification, coaching, quality, and workflow adaptation while continuing to face delivery expectations. Source
- McKinsey & Company — “The Six Strategy Roles a CEO Must Not Delegate,” October 9, 2026. McKinsey argues that CEOs remain uniquely responsible for critical strategy roles including value creation, difficult resource-reallocation choices, strategic tempo, opportunity framing, and organizational change. The article cites research associating more structured CEO strategy practices with approximately 9.5% higher five-year growth. Source
ENLOSA: Global Business Solutions
Strategy. Leadership. Execution.
enlosa@enlosa.com | +1 (877) 246-1109
#TuesdayLeadership #AILeadership #ArtificialIntelligence #DecisionMaking #DecisionRights #HumanAI #MiddleManagement #LeadershipDevelopment #CEO #COO #CFO #ExecutiveLeadership #Management #OrganizationalDesign #OperatingModel #Strategy #Execution #Accountability #AIGovernance #DigitalTransformation #Leadership #ValueCreation #OperationalExcellence #Stratactic #ENLOSA #EGBS
Leave a Reply