Key takeaways
- Coaching has a distribution problem. It has always gone to senior leaders and key talent, so most employees get little or none.
- Demand was never the issue. 55% of US workers have paid out of their own pocket for training their employer didn’t cover (University of Phoenix and Harris Poll, 2025).
- Private practice removes cost and career risk. In EasyCoach data covering about 7,000 learners, 59% came back for a second session and 23% completed five or more.
- AI doesn’t replace a great human coach. It handles practice and repetition, so human coaches can focus on judgment and nuance.
At the CLO Exchange in New York, everyone in the room either runs or is responsible for the overall L&D budget for their organization. And at the same time, more and more of their employees are going out and investing in their own coaching outside the organization.
This tells me coaching has a distribution problem, not a demand problem.
AI can make coaching and practice affordable for more and more employees. Our own EasyCoach data shows learners keep coming back. Among about 7,000 learners, 59% return for a second session and 23% have done five or more sessions.
What AI adds to coaching is that it gives employees access to a space for on-demand, private feedback and practice. Employees can rehearse a hard conversation or pitch a new business model, with an AI that plays the other person.
What I took from the CLO Exchange is that employees are already using AI for coaching and practice, at work and in their personal lives, rather than waiting for L&D.
The “AI democratizes coaching” argument is everywhere right now, and it usually leans on the same recycled data. I want to make a narrower point. We can argue about good coaching versus bad, but the bigger issue is who gets any at all. AI doesn’t make coaching much better.
What interests me is who turns up to practice once the cost and the awkwardness disappear.
Why workplace coaching has always been rationed to senior leaders
Coaching is expensive, so it has always been rationed to the top. It went to senior leaders, managers, and key talent. L&D has become far more democratized in other ways, yet in my experience coaching still reaches only a small share of employees.
L&D can’t give everyone a human coach, but AI coaching platforms are widening who can reach one.
Few employees get access to the coaching they need, and they are now paying for their own development
Employees have started solving this themselves, and the wider data points the same way, with a caveat about what it measures. A University of Phoenix and Harris Poll survey from October 2025 found that 55% of US workers have paid out of their own pocket for training their employer didn’t cover. Nearly a quarter (23%) have done so more than once. That’s training in general rather than coaching, but it shows employees will spend their own money when work doesn’t provide what they need.
Coaching from managers looks thin too. Achievers Workforce Institute surveyed 1,500 employees in June 2024, and only one in ten said they get all the coaching they need from their manager. Achievers also reports that 85% of workers say having a coach at work is important, while only 17% know how to find one. Managers may not see the gap either. In a Zenger Folkman study of 3,761 leaders, about one in four rated themselves as above-average coaches while their colleagues placed them in the bottom third (Harvard Business Review, 2016).
Why cost and career risk stop employees from asking for coaching
That employees pay their own way for development tells me the demand was always there, and access is what held it back. Cost is part of it. Asking for coaching also feels risky. It can look like admitting a weakness, and some employees worry it could count against them in their careers.
Practicing privately with AI removes that worry. That’s one reason I don’t think cost explains everything about who turns up.
This one is my own observation, and our data can’t show it. The employees practicing most aren’t the ones who’ve always had coaches. They’ve decided to address their own gaps and needs.
What EasyCoach usage data shows about AI coaching
EasyCoach is Easygenerator’s AI roleplay tool, where employees practice realistic workplace conversations with an AI that plays the other person and get personalized, structured feedback. We can see how people use it. We can’t see whether it is reaching people who were left out of coaching before, because the data doesn’t record job titles or seniority. What it can show is how people practice once they have access.
These figures run from launch to early November 2026 and cover about 7,000 learners and about 34,000 practice sessions. They include customers on trials as well as paid plans, and leave out Easygenerator’s own staff. A learner here is anyone who completed at least one practice session longer than 30 seconds.
- A typical EasyCoach practice conversation lasted about five minutes, and in a month in which they practiced, learners averaged about four sessions.
- About half of learners (49%) had repeated the same scenario at least once.
- Of all learners, 59% had come back for a second session.
- Among learners whose first session was at least 30 days earlier, 58% had returned within 30 days.
- Close to a quarter of learners (23%) had completed five or more practice sessions.
That pattern makes sense when there is nothing to schedule and no audience watching. Some customers build scenarios into multi-step programs, so we can’t separate people who come back by choice from people who were asked to.
Odido gave 1,800 customer operations agents their own AI coach. Daan Stobbe, Manager of Training and Development at Odido, describes how the agents use it.
“When one roleplay isn’t good enough, our customer service agents really want to try it again over and over. When you do an in-person session in a classroom, there’s only one or two people who can train the roleplay. Now, with EasyCoach, everyone can—without having an audience.”
You can read the Odido story.
Where AI coaching ends and human coaching begins
One person in the CLO Exchange conversation described how he splits the two. He uses AI coaching to practice the areas that matter for his role. For the more human parts of the job, where nuance and skills need discussing, he goes to his line manager. AI does the heavy lifting and the repetition, and the human supports the skills.
William Salm, a Learning and Development Specialist at Equity Trust Company, describes a similar split. “Leadership stopped being asked to run every single repetition of roleplay,” he says. “Instead, what they got to do is review the outputs, look at the data, and that’s a fraction of the time and a much higher value use of their attention.”
AI coaching doesn’t match a great human coach, though some practice beats no practice. Scale only works if it stays personal, because generic coaching isn’t really coaching. What makes mass access real is that it still feels one-to-one, and AI is starting to make it feel that way.
Some of what people bring to these tools crosses into counseling and therapy, which deserves its own article. L&D teams and managers also still aren’t aligned on what coaching, mentoring, counseling, and sponsorship are, or where each one fits.
How L&D teams can combine AI coaching and human coaching
The choice isn’t between human coaching and AI coaching. The more useful model combines both. AI provides availability, continuity, and scale, while human coaches bring the judgment, challenge, and depth that matter most in a coaching relationship.
Here’s what that looks like in practice.
- Scale. Coaching reaches many more people, not only those with the budget, seniority, or organizational sponsorship to access a human coach.
- Continuity. AI supports reflection, preparation, and follow-through between sessions, so development isn’t limited to an occasional conversation.
- Focus. Human coaches spend their time on the work that most needs their experience and judgment, while AI helps maintain momentum in between.
If I were back in a corporate role, I’d want to know where the biggest untapped pool of capability sits. It’s with the people who’ve never had any coaching, not the leaders who already do.
The aim is to make development more widely available without losing what makes human coaching valuable. The catch is quality. Access only counts if what people get is good and safe.
A call to action for CLOs, HR, and L&D leaders
CLOs, HR, and L&D leaders need to rethink who gets access to coaching. Stop reserving it for senior leaders and high-potential talent, and start making it available to the employees who have traditionally been overlooked. Combine AI-powered practice with human coaching to make development accessible to everyone.