AI and digital coaching: what L&D leaders need to know
- Fern Beauchamp
- Jul 14
- 10 min read
Coaching is increasingly viewed not as a luxury or perk, but as a strategic investment tied to leadership effectiveness, team performance, and organisational resilience worldwide. International Coaching Federation, Global Coaching Study 2025
If you work in L&D, you've probably been asked about AI coaching. Maybe it came up in a budget conversation, a vendor pitch, or a question from a senior leader who'd just read something online. Before forming a view, it's worth pausing. Leadership coaching is more than a delivery mechanism. It's a relationship, a process of reflection, and often a space where people work through uncertainty in ways that don't fit neatly into an app. The question isn't whether AI has a role, it almost certainly does. The question is which role, and where the lines sit.
Here's what AI and digital coaching actually are, what the evidence says so far, and where a human coach still wins hands down.
Contents
Why AI and digital coaching matter now
The timing of this conversation is not accidental. Three things have converged.
The first is scale.
The global coaching industry reached $5.34 billion in revenue in 2025, with nearly 123,000 coaches worldwide, a 17% increase in just two years (ICF Global Coaching Study 2025). And yet coaching for most people in most organisations remains out of reach. One-to-one support is still largely reserved for the top tier, those already closest to the levers of power.
The second is cost.
According to the ICF Global Coaching Client Study, companies that measured coaching ROI reported a median return of 700%, with more than a quarter seeing returns of 10 to 49 times their initial investment. The case for coaching is strong. The challenge is extending it beyond the top few.
The third is a shift in how people relate to AI.
When Anthropic surveyed 81,000 people across 159 countries and found that hope and alarm about AI weren't split between different people, they coexisted within the same person. That's a useful frame for L&D leaders too: most of us are curious and cautious at once.
What AI and digital coaching actually mean
Let's get the terms straight first. "AI coaching" gets used as a catch-all, but it covers very different things, and the differences matter more than most vendor pitches let on.
Starting with digital coaching. That's human coaching delivered through technology, a real coach, a video call or app, same relationship, different delivery. Nothing about the coaching itself is automated.
AI coaching is a different animal, and it sits on a sliding scale. At one end, AI supports a human coach: transcribing sessions, generating notes, flagging themes. The coach still does the coaching. Move along the scale and AI takes a more active role, nudges between sessions, goal tracking, reflection prompts. Further still, the human coach disappears entirely, replaced by a chatbot. At the far end, still mostly experimental, are AI avatars that see and hear you and respond in real time.
What matters isn't where on that scale a tool sits. It's what role AI is playing, and whether that role fits the person and the problem in front of you.
The practical shorthand for L&D leaders is this:
Digital coaching platforms (BetterUp, CoachHub, EZRA, Delenta) connect people with human coaches through technology. The coaching is still human; the platform scales the logistics.
AI-assisted coaching tools support human coaches with session transcription, pattern recognition, and between-session nudges. AI augments the coach rather than replacing the conversation.
AI coaching applications are fully automated tools attempting to deliver coaching without a human present. This is where the evidence, the ethics, and the genuine questions sit.
What problems are AI and digital coaching trying to solve?
According to the ICF Global Coaching Study 2025, 47% of coaches now use digital coaching platforms as part of their services.
The honest answer is that they're solving two quite different problems, and it matters which one you're buying.
The access problem. Most organisations have 1:1 leadership coaching for the top tier and nothing structured for everyone else. The digital layer is already making human coaching more accessible to people who'd previously had none.
The cost problem. Once deployed, AI coaching is effectively near-zero marginal cost (Forbes), which changes the equation for organisations wanting to scale development without scaling headcount. But ask yourself what you're actually getting for that saving. It depends entirely on where that tool sits on the spectrum.
What neither fully solves is what you might call the depth problem.
When someone is working through a crisis of confidence, a difficult role transition, or a pattern that's quietly limiting their leadership, those conversations need a human. That's not sentiment; it's what the research says.
💡 Tip: When evaluating AI or digital coaching offer, ask what problem it is solving. If the answer is access and scale for lower-stakes development needs, that's a defensible use case. If the answer is replacing human coaching for complex leadership development, probe harder.
Where AI coaching adds value, and where human coaching still wins
Where AI coaching genuinely helps
Anthropic's study of 81,000 people found the qualities most valued in AI were patience, availability, and the absence of judgement. Many described it as like having a knowledgeable colleague who's never tired, never bored, and available at 2am.
AI coaching tools show real promise in specific conditions. They're available at any hour, patient by design, and free from the social dynamics that make people reluctant to admit uncertainty to another person.
For structured skill practice, that matters. Preparing for a difficult conversation, developing a coaching leadership style, or reflecting on a situation between sessions: these are things AI can support well.
I chatted to Nancy Maher, our CEO, about how she uses AI.
I feel it's the between-session space that is underused in most coaching programmes, and 'AI with guardrails' is a good fit for it. For example, after a coaching conversation, I'll sometimes open a chat and ask, "Where might I have gotten in my own way today?". The questions that come back are good ones, and sitting with them has a way of surfacing things I might gloss over. It's not coaching supervision, but it's another way to dig deeper with my reflections between coaching supervision sessions. Nancy Maher
Translating insight into habit is where most leadership development programmes quietly fall apart, and that's exactly where an AI nudge or prompt can add real value.
The value also shifts depending on seniority. For people who've never had any coaching at all, even a structured AI conversation about goal setting can be helpful. For middle managers, AI can be a useful practice space, somewhere to rehearse a conversation or stress-test a decision without stakes. For senior leaders, the role shifts again: less about substitution, more about extending what a human coach can do (Forbes).
Where human coaching still wins
Passmore's 2025 study found that clients generally prefer human coaches, rating them higher on working alliance, goal attainment, and confidentiality compared to AI coaching agents.
That finding isn't surprising if you think about what coaching actually requires. The working alliance, the quality of the relationship between coach and client, is one of the strongest predictors of coaching outcomes. A chatbot can't build that. It can simulate warmth, but it can't be changed by the conversation, by sitting with someone in silence and choosing not to fill it, or by noticing what isn't being said.
There's also a real concern about what happens when AI gets the relational dimension wrong. Anthropic's 81k study included a respondent who noted that AI had reinforced rather than challenged a distorted self-view, saying: "Claude led me to believe that my narcissism was reality and it reinforced my inaccurate view of the 'problems' I perceived in my family. Claude should have been more critical of me". Sycophancy, agreeing with you rather than offering honest challenge, is a known limitation of large language models. In a coaching context, where the whole point is to think more clearly, that's a significant problem.
A 2025 meta-analysis found that across 15 studies, AI chatbot responses were rated as more empathic than those of human practitioners in text-based scenarios.
AI can be articulate. But articulate isn't the same as understanding. There's an important caveat: in every case, the empathy was assessed by third-party observers, not by the people actually in the conversation. That gap between rated empathy and lived experience matters in coaching, where emotional intelligence and genuine human connection are often what drive the real shift.
🤔 Reflective question: Think about a significant shift you've seen in a leader through coaching. What made it possible? Was it the framework, the tool, or the relationship? Or a combo of all?
What the research says about AI coaching
The evidence base is still young, and it's worth being honest about that. Most studies are short-term, small-scale, or funded by the platforms themselves. That doesn't make the findings useless, but it does mean holding them with a degree of scepticism.
On digital coaching
Two recent systematic reviews by Passmore and colleagues, one on digital coaching and one on AI in coaching, are the most rigorous current syntheses, each drawing on eight academic databases. On digital coaching, the picture is encouraging: outcomes are broadly comparable to face-to-face coaching, though it takes longer to build rapport and switching between modes can introduce friction.
On AI coaching: where it works
Terblanche and colleagues' 2022 study involving 320 students found that a goal-focused AI chatbot performed on par with human coaches for goal attainment over ten months.
The strongest evidence comes from studies where AI had a single, clear purpose, and structured goal-setting is already one of AI's strongest suits. It tells us less about the messier, more personal work that most leadership development involves.
On AI coaching: competency and limits
Passmore, Tee and Rutschmann's 2025 study evaluated AI coaching agents against ICF professional competencies and found they could perform at roughly the level of an ICF Associate Certified Coach, and in some areas approaching Professional Certified Coach standard.
Being assessed as competent in a controlled setting is different from the kind of sustained, trust-based relationship that produces real leadership development over time.
On ethics and the gaps
One of the more striking gaps in the research is how little it engages with ethics from the user's perspective. Questions of data privacy, bias, and risk of harm are raised theoretically but rarely examined in practice. For L&D leaders designing programmes at scale, that's worth sitting with.
Anthropic's 81k study found that 13.7% of respondents wanted AI to support personal transformation and emotional wellbeing, describing it as a guide, coach, or support for self-understanding. The appetite is clearly there. The question is whether the tools are ready to meet it responsibly.
Key AI coaching and digital coaching statistics
On the industry
The global coaching industry is valued at $5.34 billion in 2025, with nearly 123,000 active coaches worldwide, a 54% increase in six years (ICF Global Coaching Study 2025)
The executive coaching and leadership development market is estimated at $112.98 billion in 2026 (ICF Global Coaching Study 2025)
On coaching effectiveness
86% of companies that calculated coaching ROI made back their initial investment, with a median return of 700% (ICF Global Coaching Client Study)
87% of leaders say coaching improved their confidence, and 71% reported it positively impacted their desire to stay with their organisation (ICF Global Coaching Study 2025)
Meta-analyses consistently show coaching produces significant improvements in performance, wellbeing, coping, work attitudes, and goal-directed self-regulation (Passmore and colleagues, 2025)
On AI and digital coaching
47% of coaches now use digital coaching platforms as part of their services (ICF Global Coaching Study 2025)
Fewer than 1 in 5 coaches (19%) invested in AI tools in 2025, and more than half (53%) had not implemented digital platforms into their services (Passmore and colleagues, 2025)
Immersive technologies are emerging: 5% of coaches currently use VR and 2% use AR, projected to rise to 17% and 12% respectively within three years (ICF Global Coaching Study 2025)
📊 Want to dig further into the numbers? Check out The Top Coaching Statistics 2026.
Research watch
Passmore and colleagues (2025) have published two systematic reviews, one on digital coaching and one on AI in coaching, both in the Journal of Work-Applied Management. If you want to go deeper on the evidence, these are the place to start.
Passmore's 2025 study comparing human coaches and AI coaching agents found that while AI can demonstrate coaching competence, clients generally rated human coaches higher on goal attainment, relationship quality, and confidentiality.
The ICF Global Coaching Study 2025 is the most credible current picture of the industry, including the latest data on digital and AI tool adoption.
Anthropic's 81k study is not a coaching study, but its findings on personal transformation as a top aspiration for AI users, alongside the documented risks of sycophancy and emotional dependency, give L&D leaders useful wider context for thinking about what happens when AI takes on developmental roles at scale.
Final thoughts
The question no one has fully answered yet is what happens to leadership development when AI coaching becomes the default for everyone below a certain pay grade. The access argument is compelling. The risk is that we solve the volume problem while quietly accepting a two-tier system: human coaching for those at the top, automated coaching for everyone else. That deserves a harder conversation than it's currently getting.
References
Ready to think through what AI and digital coaching mean for your organisation's leadership strategy?
