Skip to main content
    CAREERS LOGIN

    You Chose to Implement AI. Now What?

    Written by: Thomas Deakins
    EVP, Alliances, Redwood Logistics 
    Adjunct Professor and Lecturer, Haslam College of Business at the University of Tennessee, Department of Supply Chain

    The decision has been made. Your organization is moving forward with AI. Leadership is aligned. The budget is approved. There’s urgency (and likely some pressure).

    Now comes the part that determines whether AI becomes a competitive advantage… or an expensive lesson.

    Here’s the truth:

    Deciding to implement AI is the easy part. Implementing it well is where most organizations struggle. 

    Let’s discuss what must happen next.

    1. Get Clear on Why You’re Doing This 

    Before selecting a model or signing a contract, establish clarity.

    Not:

    • "We need to be innovative."

    • "Everyone else is doing it."

    Instead, answer:

    • What specific problem are we solving?

    • Which metric are we improving?

    • What does success look like in 6 or 12 months?

    If AI cannot be tied to a measurable business outcome, you’re not leading a transformation; you’re funding an experiment.

    AI is a tool. It is not a strategy. Your strategy should already exist.

    2. Assess Your Organizational Readiness

    AI does not fix broken processes. It scales them.

    Before deployment, evaluate:

    Is your data ready?

    • Is it clean?

    • Is it standardized?

    • Is it governed?

    If your data is siloed or inconsistent, AI will simply produce flawed outcomes faster. 

    Remember, garbage in equals garbage out. Are your processes stable?

    If workflows vary by department or individual, AI will not magically align them. It requires structure.

    Are your people aligned?

    • Does leadership understand what is changing?

    • Do managers understand their responsibilities?
    • Do employees understand why this is happening?

    If the answer is unclear, pause. Readiness determines results.

    3. Pilot Before You Scale 

    One of the most common mistakes is deploying AI organization-wide too quickly.

    Instead: 

    • Establish a baseline.

    • Start small.

    • Test with defined user groups.

    • Measure performance.

    • Adjust inputs.

    • Refine processes.

    Then scale. A pilot is not about proving AI works. It’s about proving it works in your environment.

    That distinction is critical.

    4. Prioritize Change Management 

    Technology is the easy part. People are the hard part.

    AI impacts:

    • How decisions are made

    • How work is performed

    • How performance is measured

    • Sometimes, who performs the work

    Without a structured change management plan, resistance will surface.

    You need:

    • Visible executive sponsorship

    • Clear, consistent communication

    • Defined ownership

    • Adoption based milestones, not just deployment milestones

    If adoption stalls, ROI disappears. Most AI failures are adoption failures, not technical failures.

    5. Evaluate Workforce Impact Honestly 

    AI shifts responsibilities. It doesn’t simply eliminate jobs.

    Yes, some tasks will disappear. Others will evolve. New roles will emerge.

    You must determine:

    • Who requires upskilling?

    • What new competencies are needed?

    • Which roles will be redefined?

    • Where is new talent necessary?

    Equally important: how is this communicated.

    If AI feels like something happening to employees instead of with them, morale declines—and performance follows.

    Transparency builds trust.

    6. Train for Confidence, Not Compliance 

    Training is not a one-time event.

    Employees must understand:

    • How the AI works

    • When to trust it
    • When to question it
    • How it influences daily decisions

    Without confidence, employees will either avoid the system or override it and neither outcome supports transformation.

    Training should be ongoing, practical, and measured. This is not a box-checking exercise.

    7. Define ROI the Right Way

    ROI can be recognized through:

    • Faster decision cycles

    • Fewer errors

    • Improved forecast accuracy
    • Higher revenue conversion
    • Stronger customer retention

    To properly define and measure ROI, you must:

    • Establish a baseline.

    • Define expected improvement.
    • Measure at 3, 6, and 12 months.
    • If performance is not trending upward, adjust immediately.
    • Do not wait nine months to acknowledge failure.

    8. Establish Governance and Ownership 

    AI cannot be “everyone’s responsibility."

    It requires defined ownership across:

    • Model monitoring

    • Bias detection
    • Security
    • Escalation procedures
    • Performance tracking

    If no one owns it, no one manages it.

    Unmanaged AI becomes an unmanaged risk.

    The Five Factors That Matter Most

    If you remember nothing else, remember this:

    1. Define the business outcome before implementation.

    2. Ensure your data is clean and structured.

    3. Treat change management as seriously as the technology itself.

    4. Plan intentionally for workforce evolution.

    5. Measure ROI continuously—not eventually.

    Miss one, and friction increases.

    Align all five, and AI becomes leverage.

    Reality

    • AI is powerful.

    • But success will not belong to the fastest movers; it will belong to the most prepared.
    • Define the problem clearly, pilot intelligently, invest in people, and measure relentlessly.
    • AI does not create advantages on its own; leadership discipline does.

    In the next post, we’ll address another critical question:

    How do you identify the right AI use cases before spending a dollar?

    Choosing the wrong problem to solve is the fastest way to lose both budget and credibility.

    If this resonates, stay tuned. There is more coming.