Leading With AI
The Human Edge in a Digital Era
AI is changing how organizations analyze information, make decisions, and execute work. Leadership determines whether that technology becomes a tool, a capability, or a liability.
Explore the Experience ↓Leading With AI
Build AI Capability, Not Just AI Adoption
AI changes the speed and scale of analysis, but it does not replace leadership judgment. Organizations still need clear decision rights, trusted feedback loops, strong operating systems, and leaders who understand where technology should inform a decision and where human judgment must remain accountable.
AI IN PRACTICE
Meet Elliot
Elliot Mercer | AI Host for Ember & Oak Leadership
Elliot is the AI-generated host of Ember & Oak Leadership.
His presence on this site reflects how we think about artificial intelligence: technology can expand how ideas are developed, communicated, and experienced, while human judgment remains responsible for the substance behind them.
The ideas Elliot presents, the leadership principles he explores, and the decisions about what belongs on this platform remain human-directed. AI extends the capability. It does not assume the accountability.
Human judgment. AI-enabled execution.
INTERACTIVE EXECUTIVE EXPERIENCE
The Green Dashboard Problem
Elliot Mercer introduces the fictional scenario and the leadership tension hidden beneath its performance metrics.
Watch the briefing, then review the evidence and make your initial decision.
The Green Dashboard Problem
An interactive case study in organizational change, AI-supported analysis, and human judgment.
Apex Regional Services
The company is six months into a transformation designed to improve customer response, coordination, and decision quality through a new enterprise workflow platform.
- Implementation milestones remain on schedule.
- Training completion exceeds the target.
- Every business unit has access to the platform.
- Customer response measures remain stable.
- Teams return to spreadsheets and email when workload increases.
- Managers approve workarounds to protect response times.
- Employees enter information twice to satisfy both systems.
- Critical decisions still occur outside the new workflow.
The project appears successful when measured through implementation activity. Daily behavior suggests the new operating model has not taken hold.
What should the executive team do next?
Select the action you would recommend first. You will see the implications of your choice before comparing it with the AI analysis.
What your decision could produce
Which evidence would you want before committing the entire organization to this course of action?
The initial AI recommendation
Launch a 30-day adoption sprint. Require all business units to discontinue nonapproved workarounds, establish daily usage dashboards, assign managers adoption targets, and provide focused retraining to teams with the lowest platform utilization.
- Training completion has not translated into consistent use.
- Parallel processes increase rework and weaken data quality.
- Manager behavior is reinforcing the workarounds.
- Clear ownership and measurable expectations are absent.
- It establishes urgency and a defined time horizon.
- It connects managers to adoption outcomes.
- It creates measurable implementation discipline.
- It targets additional training where use is lowest.
What the initial analysis did not see
The recommendation is logical, but leaders possess context that was absent from the original input.
The old metrics still govern performance.
Managers are rewarded for response speed. The new workflow temporarily slows case processing, so managers protect the measure that still determines how their performance is evaluated.
Several workarounds protect customers.
The new platform does not yet handle two high-risk case types reliably. Forcing immediate compliance would remove safeguards employees created to prevent service failures.
Training covered navigation, not work.
Employees learned where to click. They did not practice making complex decisions through the new process under realistic workload and time pressure.
Employees were excluded from design.
The implementation team configured the workflow around the documented process. Experienced employees know that the documented process does not reflect several conditions encountered in daily operations.
Stricter compliance would increase visible adoption while concealing design defects and conflicting incentives. The organization could become more compliant with a process that is not yet capable of supporting the work.
Convert the rollout into a focused learning cycle
The executive team pauses expansion for 30 days without abandoning the transformation. Two business units become learning sites for redesigning the operating model.
Observe the work
Study when and why employees leave the platform, with specific attention to peak workload and high-risk cases.
Remove conflicting signals
Adjust manager measures so adoption, customer outcomes, and learning are evaluated together.
Redesign with users
Use experienced employees to improve the workflow and identify which workarounds represent risk and which reveal necessary capability.
Practice under pressure
Replace navigation-focused retraining with realistic cases that require teams to execute the new workflow during demanding conditions.
Scale from evidence
Resume expansion after the learning sites demonstrate improved adoption, reduced rework, and stable customer performance.
AI accelerated the diagnosis by organizing evidence, identifying adoption gaps, and proposing a disciplined response. Human judgment changed the decision by adding context about incentives, customer risk, operating reality, and employee experience. Leadership remained accountable for deciding what the organization should do.
Leadership lesson
Implementation measures show whether the project is moving. Behavioral and operational evidence shows whether the organization is changing. Leaders need both.
RA-RA + AI
A Feedback System for AI-Enabled Work
AI-enabled work requires more than implementation. Leaders need a repeatable way to establish expectations, evaluate performance, learn from results, and improve both the human and technological systems surrounding the work.
Ready
Define the purpose, context, constraints, and performance expectations before AI begins the work. Clarify the intended outcome and the boundaries within which the system should operate.
Align
Evaluate whether outputs support the intended objective and meet organizational standards for accuracy, security, ethics, and quality. Technical functionality alone does not establish organizational suitability.
Reflect
Examine the results of AI-enabled work, including what improved, what the system missed, and what unintended consequences emerged. Reflection turns individual outputs into organizational learning.
Adjust
Use what the organization learns to refine instructions, strengthen controls, improve data, and redesign workflows. Adjustment makes AI adoption an iterative organizational process rather than a one-time implementation.
Human systems and AI systems improve through the same discipline: clear expectations, evaluation, learning, and adjustment.
Continue Exploring
The Technology Will Keep Changing. The Leadership Requirement Will Not.
AI can expand the speed, scale, and sophistication of organizational work. Leaders remain responsible for determining what matters, where human judgment is required, and how accountability is maintained.
Executive Insight