Coaching deserves an evidence base. We are helping build it.
Most research into AI and coaching asks whether the AI can stand in for the coach. We are asking a different question: what a coach and an AI reach together that neither reaches alone. The studies are led by academics, and we do not fund our own verdicts. Three studies, three questions, and a way to join in.
Three studies, three different questions.
Each one asks something the other two cannot. This is where they stand today.
Can AI spot the tension a client is stuck in?
A published framework, built from repertory grids with 24 coaches, names the tensions leaders get stuck in. Our AI reads consented session transcripts and names what it sees; an independent team codes the same sessions blind. We publish how often they agree.
Does support between sessions make the learning stick?
Learning in coaching moves through discovery, application and integration (Griffiths and Campbell, 2009), and only the first happens in the hour. Coaching pairs run with support between sessions and without it, measured at the start, the end, and eight weeks after. It needs a coach community large enough to run both arms at once.
Does the coaching relationship get stronger or weaker?
The question coaches ask first, and the one we would rather answer with evidence than reassurance. The working alliance measured from both sides, where AI is part of the practice and where it is not. This one is early, and coaches who want to shape it with us are invited.
We publish what we find. A study that can only confirm us is not a study.
These are our three questions. If you are sitting on a fourth, we would rather study it than guess at it.
Bring us a questionWorking with Dr Roger Maitland.
Roger’s framework, published this year in the International Journal of Evidence Based Coaching and Mentoring, with support from an EMCC Global grant, is the vocabulary our AI uses to read a session. His paper invites exactly the AI validation we are proposing, and we hold the one thing it needs: consented, longitudinal sessions from real engagements, with the coach in the loop.
He leads the study and holds first authorship. We provide the platform and the data, and we do not touch the conclusions.
The three tensions
Maitland, Anderson-Terry and Moore, 2026. International Journal of Evidence Based Coaching and Mentoring →
The money and the verdicts stay out of our hands.
We want the answer to come out a certain way. That is exactly why none of it is ours to decide.
- The academics lead. First authorship and the analysis are theirs; every interest, including ours, is declared.
- Registered before results. Question, measures and analysis locked before any data is looked at.
- Coded blind, by outsiders. Nobody coding sessions holds a financial interest in CoachNova.
- Consent is separate. Research is its own opt in for every client, not bundled into the terms.
We are looking for the people who make this rigorous.
Researchers with a neighbouring question. Coach communities who can carry a cohort. Coaches who want in when a study opens. We hold what is hard to assemble, consented longitudinal sessions from real engagements, and you set the method.
Follow our R&D.
Every stage goes out in the newsletter, method first, results when published.