What Is AI Coaching, and Does It Actually Work?
AI coaching refers to development support delivered through a chatbot or AI-driven platform rather than, or alongside, a human coach, typically helping someone set goals, reflect on progress, and stay accountable over time. The evidence on how well it works is still developing. An early trial found AI could match human coaching on goal attainment in a narrow context, while the first direct head-to-head trial found human coaching clearly outperformed it. Read together, the research points to AI coaching adding real value in a specific supporting role, not as a replacement for a skilled human coach.
What the Research Actually Found
Terblanche, Molyn, de Haan, and Nilsson (2022) conducted two separate but equivalent longitudinal randomized controlled trials over a ten-month period. One compared human coaching against a control group, and the other compared an AI chatbot coach against a control group. Both formats produced significant increases in goal attainment relative to their respective controls, and comparing across the two trials, the AI coach matched the humans on goal attainment.
The researchers attributed this to the AI coach's strict, consistent adherence to goal-attainment theory and structured methodology, which appeared to compensate for its lack of a skilled human coach's emotional intelligence. Two things are worth being precise about, though. This was a comparison across two separate trials rather than a head-to-head test, and the effect was specific to structured goal attainment rather than to coaching outcomes more broadly.
A more recent study put the question to a direct test, and the result was different. de Haan, Terblanche, and Nowack (2026) ran a three-arm randomized controlled trial with 114 senior leaders, comparing human coaching, AI chatbot coaching, and a control group directly for the first time. Human coaching produced significant improvements in goal attainment, stress reduction, and coaching effectiveness, while AI coaching showed no significant improvement over the control group on the primary outcomes.
Two further findings stand out. Dropout in the AI group was far higher than in the human group, and several participants asked to switch to a human coach partway through. The authors note that their results do not support the earlier 2022 finding, and they point to the coaching relationship, the give-and-take between coach and coachee, as what AI struggled to reproduce.
Where AI Coaching's Strength Actually Lies
Dr. Luke Treglown, Director of AI and Assessment R&D at Deeper Signals, frames this result in a way that clarifies what it does and does not mean. AI coaching's core strength is not that it out-coaches a skilled human. It is that it solves a genuine scalability problem that human coaching, by its nature, cannot solve on its own.
Human coaching is time-limited, relatively expensive, intermittent, and dependent on scheduling. AI coaching can be always available, continuous, low-cost at scale, and integrated directly into someone's daily workflow. This makes it particularly well-suited to the specific mechanism Terblanche et al. (2022) found effective: regular check-ins, consistent reinforcement of goals, and structured accountability over an extended period, which is precisely the kind of consistent, structured engagement their AI coach delivered.
Where the Evidence Is More Limited
Dr. Treglown is also candid about where AI coaching's evidence base is thinner. AI systems can default toward overly solution-focused responses rather than sitting with difficult emotions or facilitating the kind of deeper exploration a skilled human coach knows when to slow down for. They may not always align cleanly with best-practice therapeutic or coaching frameworks that depend on relational nuance rather than structured consistency.
The honest summary is that AI coaching's evidence is strongest for structured, goal-oriented outcomes, and thinner for the deeper relational and emotional dimensions of coaching.
Assessment Data Makes AI Coaching More Personal
AI coaching becomes considerably more useful when it is paired with existing assessment data rather than operating as a generic chatbot. AI can translate a person's assessment results into specific development goals, align those goals with stated career aspirations, generate a structured coaching plan from that combination, and adapt the plan over time as the person makes progress or their circumstances change. This creates a development experience that responds to the individual rather than delivering the same generic guidance to everyone.
The Blended Model: Where Humans and AI Each Add the Most Value
The evidence points toward a blended model rather than a choice between human and AI coaching. Humans are best positioned to define what actually matters to a person, establish meaningful goals and direction, build the trust that makes coaching work at all, and navigate genuinely complex emotional conversations. AI is best positioned to maintain rhythm and consistency between those human touchpoints, provide ongoing accountability, and make coaching accessible and continuous in a way that scheduled human sessions cannot match on their own.
In practice, this suggests humans should lead at the start of a coaching relationship, building trust and establishing goals and meaning, while AI supports the ongoing middle of the journey through check-ins, nudges, reflection prompts, and consistent accountability. One further, specific advantage of AI in this space is that some people feel more comfortable asking a naive question or admitting uncertainty to an AI system than to a human coach, which can encourage more frequent and more honest engagement, particularly in earlier-stage or lower-stakes development conversations.
How Deeper Signals Approaches This
At Deeper Signals, Dynamo and Sola reflect this blended model directly. DynaMo is a digital development tool built on the science of the Core Drivers Diagnostic, curating personalized goals and weekly activities based on a person's actual assessment profile rather than generic content, and reinforcing progress consistently over time in exactly the way the research suggests AI coaching does best. Sola complements this by giving people an always-available way to ask questions about their own results and get situational guidance grounded in their actual assessment data, extending support into the moments between formal coaching touchpoints.
For the deeper, relational side of coaching, Deeper Signals offers Core Coaching, pairing each participant's assessment results with a certified human coach who builds a development plan around their specific transition. This reflects the blended model this research supports: human coaches lead on trust, direction, and complex conversations, while AI tools like DynaMo and Sola maintain the ongoing rhythm, accountability, and accessibility that make development stick between sessions.
Frequently Asked Questions
Is there real evidence that AI coaching works, or is this mostly speculation?
There is genuine but mixed evidence. Terblanche et al. (2022) found an AI chatbot coach could match human coaching on goal attainment in a narrow, structured context. But a more recent direct trial, de Haan, Terblanche, and Nowack (2026), found human coaching outperformed AI, which did not beat the control group.
Does AI coaching work as well as human coaching for everything?
No. The strongest evidence for AI coaching is in structured, goal-oriented outcomes like goal attainment. The evidence is thinner for deeper relational and emotional dimensions of coaching, where a skilled human coach's judgment about when to challenge, slow down, or sit with difficult emotion remains harder to replicate.
Can AI coaching be personalized, or is it generic advice?
AI coaching becomes meaningfully more personalized when it is paired with an individual's actual assessment data. This allows the AI to generate development goals and coaching plans tailored to that person's specific profile and aspirations, rather than delivering generic, one-size-fits-all guidance.
Should organizations replace human coaches with AI coaching?
The evidence does not support full replacement. It supports a blended model, where human coaches lead on trust, goal-setting, and complex emotional conversations, and AI provides the ongoing consistency, accountability, and accessibility that sustains progress between human sessions.
Why might someone prefer talking to an AI coach over a human one?
Some people feel more comfortable asking basic questions or admitting uncertainty to an AI system without fear of judgment, which can lead to more frequent and more honest engagement, particularly for early-stage development or lower-stakes coaching conversations.
Last reviewed by Dr. Reece Akhtar — June 2026
References
de Haan, E., Terblanche, N., & Nowack, K. (2026). A randomised controlled comparison of the effectiveness of human and AI chatbot coaching with goal attainment, wellbeing and self-efficacy. Human Resource Development International, 29(4), 754–783. https://doi.org/10.1080/13678868.2026.2633990
Terblanche, N., Molyn, J., de Haan, E., & Nilsson, V. O. (2022). Comparing artificial intelligence and human coaching goal attainment efficacy. PLOS ONE, 17(6), e0270255. https://doi.org/10.1371/journal.pone.0270255
