Talk Less, Do More

How role-play and AI transformed a struggling trainer.

At 4:55 p.m. on a Friday, a message from my boss popped up: “Need to chat.”

My mind immediately categorized the conversation as “bad.”

I called him, and my intuition was right.

“We may need to let Mark go,” he said.

Mark was a longstanding sales trainer, well-liked and respected. But his cohorts consistently finished last in sales metrics. His evaluations were mediocre. The Operations team felt his classes left new hires underprepared. His confidence was slipping, and he blamed recruiting and bad luck.

The human cost was real, but so was the business impact. An underperforming trainer meant unprepared employees and lost revenue.

I made a bet with my boss: “Give me one month. If Mark’s next class isn’t in the top 10 percent of performers, we’ll move forward.”

The Curse of Knowledge at Work

When I observed Mark’s class, the problem was clear. The very knowledge that made him an expert had become his greatest obstacle. He had fallen victim to the curse of knowledge, which is the bias where an expert forgets what it is like to be a beginner (Newton, 1990).

Early in his career, Mark couldn’t lecture on theory, so he relied on role-plays and simulations.

He focused on doing the job, not talking about it. Years later, armed with curriculum expertise, he shifted into lecture mode. In the process, he had stopped being a coach and become a lecturer.

This is a common trap. Talking feels more productive, and it is more comfortable than facilitating unpredictable role-plays. Yet this runs counter to learning science. Research shows learners retain 75 to 90 percent of what they do, compared to just 10 to 30 percent of what they read or hear (eLearning Industry, 2025).

From Knowledge Transfer to Skill Certification

The solution wasn’t to lecture Mark on lecturing. It was to change the system.

We shifted the goalpost from information delivery to skill certification. Before trainees could advance in training—meaning before they were allowed to take live calls with customers—they had to pass a rigorous role-play with the training manager. About half failed on the first attempt.

The trainer’s role was to prepare them to succeed, ensuring they could demonstrate competence in the most critical parts of the job. The impact was immediate. The assessments forced practice, not just participation. We restructured class time: more role-plays, peer feedback in groups of three, and live demonstrations from top performers.

By the third day, the room had shifted. Trainees leaned into role-plays, laughed when they stumbled, and sharpened one another’s skills. Mark wasn’t lecturing anymore. He was orchestrating.

His next class became the top-performing cohort in the company.

Where AI Fits

What turned Mark’s performance around was the virtuous cycle of simulation, feedback, and assessment. The challenge is that most trainers won’t sustain that intensity. It is exhausting, and learners often resist.

This is where AI makes a difference.

We used an AI-powered role-play tool trained on our own call transcripts, objection libraries, and top-performer examples. It simulated realistic customer conversations and gave instant feedback on key criteria we defined, such as handling objections and reinforcing value before discussing price. We validated its accuracy by comparing its feedback with trainer evaluations and confirming that higher AI scores aligned with stronger live performance.

The results mirror what research is finding at scale. Walmart found that employees using AI-powered simulations cut training time by 95 percent and improved performance by 15 percent (eLearning Industry, 2025). IBM saw a 40 percent increase in employees demonstrating mastery of key skills after deploying AI-driven learning paths (SuperAGI, n.d.).

AI doesn’t replace the human element that energizes a classroom. It amplifies what works: structured simulation, immediate feedback, and consistent reinforcement, so trainers can focus on mentoring, inspiring, and connecting.

Lessons for Talent Development

Mark’s story carries three lessons that apply to every organization:

  1. Assess skills, not knowledge. Redefine success by requiring employees to demonstrate capability, not recall theory.
  2. Target the hardest moments. Focus practice on the make-or-break parts of the job: objections, negotiations, pricing conversations.
  3. Leverage AI for scale. Use AI to automate role-play, feedback, and repetition, freeing trainers for high-value coaching.

The Callback

One month later, at 4:55 p.m. on a Friday, another message from my boss appeared: “Call me.”

My stomach sank.

But this time, his words were different: “I just want to congratulate you on your work with Mark. I’m impressed.”

A simple shift from knowledge transfer to skill application had turned a termination into a transformation. With AI, organizations can deliver that same transformation across their entire workforce.

References

eLearning Industry. (2025, July 27). 6 companies that have successfully implemented AI in corporate training programs.

https://elearningindustry.com/case-studies-successful-ai-adoption-in-corporate-training

InteDashboard. (n.d.). 5 reasons why immediate feedback is an effective teaching strategy.

https://www.blog.intedashboard.com/blogs/tbl-learning/immediate-feedback

Newton, E. L. (1990). The rocky road from actions to intentions [Unpublished doctoral dissertation]. Stanford University.SuperAGI. (n.d.). AI vs. traditional methods: Comparing the effectiveness of AI training content generators in corporate learning.

David Smailes
David Smailes is an executive coach with CxO Coaching. He helps senior leaders and organizations accelerate growth. He speaks on AI’s role in leadership and learning and has led global teams at Fortune 500s and startups.