THE CASE FOR AI COACHING
Most arguments for AI coaching are opinions. This page isn’t. It’s the cost data, the published research, and the specific architecture we’ve built to close the gap that’s left.
The debate about whether AI coaching matches human coaching usually skips a more basic fact: most employees were never in line for human coaching to begin with.
Drag the slider. The traditional estimate is built from a realistic seniority mix and published market coaching rates, not a hand-picked number. Full methodology below.
This is our own illustrative model, not a single cited study — built by blending published market coaching rates across a typical enterprise seniority mix. We're showing it so you can check our assumptions, not to ask you to take a number on faith.
| Tier | Share of headcount | Rate used | Source |
|---|---|---|---|
| Senior leadership / C-suite | 3% | £20,000 / person / year | Enterprise executive coaching spend, converted from published USD market data6 |
| Managers & directors | 20% | £2,750 / person / year | Group coaching programs, converted from published USD market data7 |
| Everyone else | 77% | £550 / person / year | Estimated — most organizations don't currently fund coaching at this tier at all |
We’re not going to tell you the research is unanimous. It isn’t. Here’s what it actually says:
The Conference Board found AI can already perform up to 90% of day-to-day coaching functions — summarizing, paraphrasing, active listening, structured goal work.
When assessed against the International Coaching Federation's own competency framework, an AI coaching tool met the standard for Associate Certified Coach (ACC).
In a longitudinal randomized controlled trial run over 10 months, an AI chatbot coach was as effective as human coaches at helping clients reach their goals.
A 2025 study found no significant difference in the working alliance — the trust and rapport between coach and client — built with an AI coach versus a human coach in a single session.
A separate small-scale study found clients rated human coaches somewhat higher than an AI coaching agent on working alliance and goal commitment after one session. The authors flagged limited sample size as a real constraint on the finding.
Most big coaching platforms bolted AI onto the outside of a human coaching model — scheduling, note-taking, session insights. The coaching itself is still delivered by a human, on a human’s calendar, at a human’s price.
Closing the remaining gap between AI and human coaching isn’t a prompt-engineering problem. It’s an architecture problem — and it’s what we’ve spent most of our engineering effort on.
Personalises how each coach shows up to each person over time, so the relationship deepens the way a real coaching relationship does — rather than resetting every session.
We deconstructed a coaching conversation into 72 individually measure parameters. Every coaching conversation is scored against this matrix, so "good coaching" is a measured standard, not a hope.
Our autonomous improvement loop that uses those quality scores to keep making the coaching sharper, without waiting on a manual retrain cycle.
Bring this to your organiszation and we’ll walk your own headcount and seniority mix through the model above.
Sources
1. The Conference Board, “AI Can Provide 90% of Career Coaching,” 2025
2. Range consistent across GBP, USD, and AUD-denominated market data, 2025–2026
3. Luisa Zhou / ICF client demographic data
4. Gallup / SHRM, manager impact on engagement variance
5. Gallup, “Give Up Bossing, Take Up Coaching,” 2024–2026
6. ICF / Careerminds, enterprise annual coaching spend per executive
7. Leaders ADAPT, group coaching program pricing, 2025
8. Terblanche et al., “Comparing AI and human coaching goal attainment efficacy,” PLOS ONE
9. Barger, “AI vs. human coaches: working alliance in a single session,” Frontiers in Psychology, 2025
10. Journal of Work-Applied Management, human coaches vs AI coaching agents, 2025
