When I worked for Avon Products, we had a terrific IT director named Larry. Larry would often say: “It’s great to have a lot of activity on the field, but I’ve got to see the points on the board.” In other words, activity isn’t the same as value. Good leaders don’t reward motion. They reward meaningful outcomes.

Larry’s words came back to me recently as I started reading about two of AI’s newest buzzwords: token-maxxing and model-maxxing.

Apparently, we’ve entered the age of maximizing everything. We maximize protein intake, sleep scores, daily steps, and fiber. Now we’re maximizing AI.

First came token-maxxing, the idea that organizations should encourage employees to generate as many AI prompts and consume as many tokens as possible. More usage meant more adoption, and more adoption was assumed to mean more value.

When that logic began to show its limits, along came model-maxxing. Rather than simply maximizing usage, organizations began optimizing which AI model employees should use for which task. Straightforward work is routed to smaller, less expensive models while more complex work is assigned to frontier models that cost more but promise better results.

Neither approach is inherently wrong. In fact, both may reduce costs and improve efficiency. The problem is assuming they’re strategies when they’re not. They’re optimization tactics.

An AI strategy would answer a different question altogether: How will AI help us create greater value for customers, improve organizational performance, and strengthen our competitive advantage?

Those aren’t technology questions. They’re leadership questions.

Organizations now have dashboards filled with token counts, usage statistics, model-routing efficiencies, and cost comparisons. Those metrics may tell us whether AI is being used efficiently. They tell us very little about whether the organization itself is becoming more effective.

Why Smart Leaders Fell Into This Trap

It’s worth understanding why they did it. Usage is easy to measure.

Executives needed evidence that their AI investments were gaining traction. Boards wanted proof that expensive licenses weren’t sitting idle. Software vendors naturally highlighted engagement metrics because they demonstrated adoption.

The problem arose when organizations began confusing evidence of activity with evidence of progress.

Management has wrestled with this challenge for decades. Measure call volume and people make more calls. Measure hours worked and people stay later. Measure emails sent and inboxes fill up.

AI hasn’t changed this management principle. It has simply given it a new vocabulary.

The organizations making the most progress aren’t asking, “How many prompts did employees generate this month?”

They’re asking, “How has AI improved the quality of our decisions, our customer experience, our speed, our innovation, and our ability to compete?”

AI Isn’t a One-Time Implementation

Many organizations still approach AI like previous technology rollouts.

• Purchase software.
• Train employees.
• Track adoption.
• Move on.

AI doesn’t work that way.

Unlike most enterprise technologies, AI continues to evolve almost daily. New capabilities emerge, models improve, workflows change, and employee expectations shift just as quickly. AI isn’t a one-time implementation. It’s becoming an ongoing condition of work.

Organizations don’t just need AI capability. They need the capacity to continuously adapt alongside it.

That distinction matters because organizations that treat AI as a project often stop once people begin using the tools. Organizations that treat AI as an organizational capability continue redesigning work, building new habits, strengthening judgment, and improving how people collaborate with AI over time.

Adoption Is Only the Beginning

One of the biggest misconceptions about AI is that adoption equals success.

Adoption simply means people are using the technology.

AI implementation progresses through three stages.

The first is Adoption. People have access to AI, experiment with it, and begin learning basic skills.

The second is Integration. AI becomes embedded within everyday workflows. Teams establish shared expectations about when to use AI, when human judgment takes precedence, and how work should flow differently because AI is now part of the process.

The third is Amplification. This is where AI begins producing measurable improvements in organizational performance because people have redesigned how work gets done rather than simply accelerating existing processes.

Surface Adoption Isn’t Transformation

One of the more insightful observations I’ve seen is Gartner’s description of surface adoption.

Employees use AI. They complete more prompts. They become more comfortable with the technology. But they don’t fundamentally change how they work.

That’s exactly what happens when organizations chase usage metrics.

People write the same reports. Attend the same meetings. Follow the same approval processes. Complete the same manual handoffs. AI simply helps them do those activities a little faster.

That’s useful. It just isn’t transformational.

The organizations realizing the greatest value from AI aren’t asking, “How can we use AI in this process?”

They’re asking a much better question: If we were designing this process today, knowing what AI can do, would we build it this way?

Instead of automating individual tasks, leaders begin redesigning entire workflows.

• Which approvals exist only because information once moved slowly?
• Which reports no longer need to exist?
• Which meetings disappear?
• Where should human judgment become even more important?
• Where can AI eliminate repetitive work so people spend more time solving problems, coaching others, and building relationships?

Those conversations create business value. More prompts don’t.

The Real Competitive Advantage Is Human Judgment

Ironically, AI is making uniquely human capabilities more valuable, not less. The organizations pulling ahead aren’t necessarily using the most sophisticated models.

They’re developing better judgment.

One of the biggest shifts leaders need to make is moving employees beyond prompting and toward problem framing.

Before asking AI for an answer, teams should first ask themselves:

• What problem are we actually trying to solve?
• What context does AI need that it doesn’t already have?
• What would success look like?

Those questions determine the quality of AI’s output far more than clever prompting techniques. That’s why AI fluency isn’t simply knowing how to use the technology. It’s knowing when to trust it, when to verify it, and when human judgment should override it entirely.

Trust Is the Missing Multiplier

None of this works without trust. Not trust in the technology alone, but trust in leadership and one another. Trust that asking questions won’t be interpreted as resistance. Trust that challenging AI’s recommendations won’t be punished. Trust that experimentation is expected because learning is too.

Organizations often assume employees resist AI because they’re afraid of change. In my experience, uncertainty is far more common than resistance. People want clarity and boundaries; they want to know what’s expected of them.

They want leaders who model curiosity instead of pretending to have all the answers. Leaders who openly share their own experiences with AI, including where it succeeded and where it fell short, build far more credibility than leaders who simply mandate adoption from the front of the room.

That kind of psychological safety encourages people to experiment, share lessons learned, and improve together. Those behaviors are what ultimately transform AI from an interesting tool into an organizational capability.

Stop Measuring Activity. Start Measuring Performance.

Leaders shouldn’t focus on token-maxxing, model-maxxing, or whatever the next optimization trend happens to be. They should ask whether AI is improving decision quality, shortening cycle times, reducing risk, strengthening customer relationships, increasing innovation, and enabling employees to spend more time where human judgment creates the greatest value.

And leaders should listen to Larry: Activity is easy to generate. Value requires intention.

AI gives organizations extraordinary new capabilities, but technology alone doesn’t create competitive advantage. Organizations do.

The companies that outperform over the next decade won’t necessarily be the ones generating the most prompts or using the most advanced models. They’ll be the ones that build leaders who communicate clearly, redesign workflows thoughtfully, cultivate cultures of trust and continuous learning, and help people use AI to make better decisions rather than simply faster ones.

That’s the work we do at CRE Solutions through The Human Shift — building leaders and teams who create real value with AI, not just more activity around it