I’ll start with an opinion that might annoy a few vendors: most of what’s being sold as “AI in leadership development” in 2026 is a chatbot stapled to a content library. The marketing decks are gorgeous. The outcomes are thin.
I’m the CEO of Risely, so I have skin in this game. We build an AI coach (his name is Merlin) that sits inside Slack and Microsoft Teams and coaches managers through their actual work, not hypothetical case studies. I’ve personally coached over 300 managers, and before this I spent years earlier in my career running corporate strategy inside a large multinational, which means I lived the frozen middle from the inside. I watched smart, well-meaning managers get ground down by quarterly chaos, unclear expectations, and leadership development that showed up once a year and then disappeared.
So when people ask me what AI in leadership development actually does, I try to answer the way I’d want to be answered if I were on the buying side of the table. That’s what this post is.
The Old Leadership Development Model Is Breaking
Let’s be direct about what’s failing.
Workshop-heavy programs are the default for most companies, and they’ve been the default for thirty years. A two-day offsite, a binder, a few role plays, and back to work. The research on this is unkind. A long-running study from the Center for Creative Leadership found that only about 11% of organizations feel confident their leadership pipeline can fill critical roles. Meanwhile, Harvard Business Review has repeatedly reported that companies spend roughly $14 billion a year on leadership development and most of it doesn’t stick.
The second failure is that serious coaching is reserved for the C-suite. Executive coaches charge $500 to $1,500 an hour. If you’re a frontline manager running a team of eight, you’re not getting coaching. You’re getting an LMS link on Tuesday.
The third failure is the one I care about most: the gap between moments. A manager learns something in a workshop on Thursday. On Monday morning, she has to confront an underperforming team member, and there is nobody to help her rehearse the conversation. By Tuesday, she’s forgotten half of what the workshop covered. That gap, between the learning and the moment the learning is needed, is where most leadership development dies.
AI in leadership development is interesting to me because it’s the first thing I’ve seen that can actually close that gap. But only if you build it the right way.
What AI in Leadership Development Actually Does (And Doesn’t Do)
I want to be precise here because the category is crowded with promises.
What AI coaching does well in 2026:
It personalizes at scale. A good AI coach can remember what a manager is working on, who her team is, what her last 1:1 sounded like, and meet her where she actually is. Human coaches do this too, but they cost 300 times more and can’t be reached at 11:47 PM the night before a tough conversation.
It provides practice at volume. Managers can rehearse difficult conversations, delegation scripts, and feedback moments as many times as they need, without the awkwardness of asking a colleague to role-play with them.
It’s there in the moment. This is the big one. Every manager I’ve coached has had situations that came up at 9:15 AM with no warning. A human coach wasn’t available. A workshop from three months ago wasn’t useful. An AI coach in your Slack or Teams, asking the right questions, actually is.
What it doesn’t do (and if a vendor tells you otherwise, push back):
It doesn’t replace the human relationships a leader has with her team. No amount of AI changes the fact that trust is built in real conversations with real people.
It doesn’t fix broken culture. If your org punishes honest feedback or rewards politics, an AI coach will be used twice and then abandoned. Technology can’t patch a cultural problem.
It doesn’t make up for a bad manager who refuses to be coached. The coachee still has to show up. AI makes showing up easier, but it doesn’t make showing up happen.
Honest take from the builder’s chair: AI in leadership development is a practice layer, not a replacement layer. If you think of it that way, you’ll buy well. If you think of it as “we can fire our coaches now,” you’ll be disappointed.
Five Places AI Coaching Actually Works
These aren’t hypothetical. These are the five use cases where I see managers come back to Merlin over and over again.
1. Rehearsing a hard conversation before it happens. A manager has to tell a senior engineer his work is slipping. She has never had this conversation before. She opens Slack, tells Merlin the situation, and spends ten minutes walking through how she wants to open the conversation, what she’ll say if he gets defensive, and what her non-negotiables are. By the time the actual meeting happens, she’s rehearsed three versions and picked the one that fits her voice. That’s not a product demo. That’s what our users do on an average Tuesday.
2. Debriefing a meeting that went sideways. A team standup goes badly. Two people interrupt each other, a decision gets deferred, and the manager leaves feeling frustrated. Thirty minutes later she types a few sentences about what happened. Merlin asks her what she thinks the root cause was, what she could have done differently, and what she wants to do before the next standup. It’s a reflection partner. That’s it. But reflection partners are rare, and most managers don’t have one.
3. Getting a second opinion on written communication. A manager drafts a Slack message to a peer she’s in conflict with. She pastes it into Merlin and asks, “does this sound defensive?” Merlin points out two phrases that read as accusatory and suggests softer framings. She sends the better version. This is small and unglamorous and it happens dozens of times a day across our user base.
4. Scenario-based skill practice. A new manager knows she’s supposed to delegate more, but she’s never been taught how. She asks Merlin to walk her through a delegation conversation with a specific team member. Merlin plays the team member. She practices. She gets feedback on where her delegation was vague. Try that with a workshop binder.
5. Pattern recognition across 1:1 notes. This is the sleeper use case. A manager keeps short notes from her weekly 1:1s. Over eight weeks, Merlin can spot that two of her direct reports have mentioned workload three times each, and she hasn’t addressed it. Pattern recognition at this granularity is almost impossible for a human to do manually across a team of eight.
Notice what’s missing from this list: no “AI generates your entire leadership curriculum,” no “AI predicts who your next leaders will be,” no “AI replaces your coaches.” Those are marketing claims. The five above are the real reasons managers come back.
What to Look For in an AI Leadership Development Platform
If you’re evaluating platforms (and you should be, there are a lot of them now), here’s the checklist I’d use. I’m biased because I built one of these, but I’ll try to give you criteria that would apply even if you picked a competitor.
Does it work where work happens? If your managers have to open a separate app to get coaching, they won’t. Leaders don’t adopt another tab. The coaching has to show up inside Slack, Microsoft Teams, or wherever the team already lives. We built Merlin to run natively inside Slack and Teams for exactly this reason. Full coaching sessions, not just nudges. If a platform requires a new login every time, that’s a red flag.
Daily reps, not monthly sessions. Leadership is a skill. Skills improve through repetition. If the platform’s engagement model is “log in for a one-hour session twice a month,” it’s a digitized version of executive coaching, and it will have the same reach problems. You want something that can deliver small, frequent touchpoints. We see 73% high engagement with daily nudges on our platform, and that compounds into about 26% average skill improvement in 12 weeks.
Skill-level tracking, not completion tracking. Most LMS-style platforms report on completion rates. That tells you nothing. You want a platform that can tell you which specific skills a manager is improving on, and ideally one that maps those skills against a real framework. Our coverage is 83 skills across 1,000+ O*NET occupations, but the specific number matters less than the principle: you should be able to see movement at the skill level.
Confidentiality for the coachee. If a manager thinks her coaching conversations are being forwarded to HR, she will never be honest with the coach, and the coach is useless. Self-driven coaching conversations should be private. Assigned development plans can have aggregated engagement data shared with L&D (we do this), but the content of conversations should stay between the coachee and the coach. Ask every vendor directly: who sees the conversation logs?
An audit trail for L&D. This is the mirror of the previous point. L&D still needs to prove the program is working. You want aggregate dashboards, skill movement at the cohort level, engagement data, and ROI signals, without exposing individual conversation content. The good platforms do both.
If you want to go deeper on how different platforms stack up on these criteria, we’ve put together a detailed guide at the best AI coaching platforms for leadership development. I tried to keep it honest even about our own gaps.
Why We Built Risely This Way
We want to be direct about the choices we made and why, because they weren’t obvious at the time and they shape how Risely works today.
We put Merlin inside Slack and Teams natively because we’ve watched, across years of working inside large multinationals, how many “leadership tools” got rolled out with great fanfare and then died because nobody wanted to open another app. Managers live in Slack or Teams all day. If coaching isn’t there, it isn’t anywhere. So we built full coaching sessions inside those tools, not just notification nudges.
We built Merlin to support 40 languages in both voice and chat because the idea that leadership coaching is an English-only privilege is something we refuse to accept. We’ve coached managers in India, in Latin America, in Southeast Asia, in Europe. Their insights were often sharper than the ones we heard in New York boardrooms. But they were systematically locked out of coaching because of language. If a manager in Jakarta can get the same quality of reflection as a manager in Chicago, that changes who gets to lead.
We built the platform around daily nudges and short touchpoints because we’ve never seen theory beat practice. A manager who rehearses a difficult conversation three times before the meeting is a better manager than one who read a book about difficult conversations last month. Every product decision we’ve made has been in service of “more reps, more moments, more practice.”
We’ve worked with 5,000+ managers across 40+ organizations so far, and the feedback that matters most to us isn’t the NPS score. It’s when a manager messages us directly and says, “I used Merlin to prep for a conversation I was dreading, and it went better than any conversation I’ve had with that person.” That’s the product working the way it’s supposed to.
The Limits of AI Coaching (Be Honest About This)
I’d be a bad advocate for this category if I didn’t tell you where it stops.
AI can’t replace the human relationships a leader has with her team. If a manager uses Merlin as a substitute for actually talking to her people, she’s doing it wrong. The coaching is supposed to make the real conversations better, not replace them.
AI won’t fix a toxic culture. If the problem is that your company punishes honesty, no coaching platform will save you. You need a different intervention, and probably a different leader at the top.
AI doesn’t work if the leader isn’t willing to be coached. This sounds obvious but it’s worth saying. A manager who thinks she has nothing to learn will use any coaching platform for two weeks and then stop. Willingness is a prerequisite we can’t manufacture.
AI is not a substitute for feedback from real direct reports. Pattern recognition across notes is useful, but it’s not the same as hearing from your team. The best managers I know use AI coaching alongside regular 360-style input from the people they actually work with. Speaking of which, if your team doesn’t have a structured way of giving this kind of feedback, we have a free constructive feedback assessment that a lot of managers use to start the conversation.
If anyone in this category tells you AI is a replacement for all of the above, walk away from the conversation. It isn’t, and pretending otherwise is how the category gets a bad reputation.
Try It Yourself
The honest answer to “does AI in leadership development work in 2026” is: yes, when it’s built around daily practice inside the tools managers already use, with real skill tracking and real privacy. It doesn’t work as a glorified LMS or a once-a-month session.
If you want to see what I mean, you can try Merlin for free. No credit card, no sales call. Bring a real situation you’re facing this week and see how the conversation goes. That’s the only test that matters. If you’re evaluating for your organization, you can also look at how Risely handles leadership development at scale or take the leadership skills assessment to see where your own skills stand today.
Frequently Asked Questions
Is AI in leadership development going to replace human coaches?
No, and I say that as the CEO of an AI coaching company. Human coaches do things AI can’t, especially around deep emotional work, career transitions, and relationship trust. What AI does is make daily coaching available to managers who would never have access to a human coach. The two work together. Executives still get human coaches. Frontline managers finally get coached at all.
How is this different from a smart chatbot or using ChatGPT for leadership advice?
A general-purpose AI doesn’t know your role, your team, your company’s context, or your development goals. It doesn’t track your skill progression over time. It doesn’t sit inside the tools you already use. And critically, it isn’t trained specifically on coaching methodology, which is a real discipline with real techniques. A purpose-built AI coach like Merlin is trained on coaching craft and anchored to a skill framework that tracks real movement. You can also see the difference yourself with our coaching skills assessment.
What about privacy? Who can see what my managers say to the AI coach?
On Risely, self-driven coaching conversations are fully private to the coachee. For assigned coaching plans, L&D sees aggregate engagement signals and session summaries, not the actual conversation content. We don’t use user conversations to train models. Ask every vendor this question directly. If they can’t answer it clearly, that’s your answer.
How long before I see results from AI coaching?
In our data, the average manager sees about a 26% improvement in targeted skills within 12 weeks, and the gains compound the longer they stay with it. That said, a manager who uses the platform for two weeks and then drops off won’t see anything. Consistency is the variable that matters most. One of the best places to start is a specific skill like delegation, because it’s concrete, it’s measurable, and most managers know they need to get better at it.
Should we run a pilot before buying?
Yes, always. I’d actively distrust any vendor who pushed you to skip a pilot. Start with one team, set a clear hypothesis, measure what you care about (skill movement, manager confidence, specific behaviors), and decide after 8 to 12 weeks. That’s enough time to see real movement without committing the whole organization to something that hasn’t proven itself yet.