Nobody argued about the AI

This month I sat down with three of the most established coaches I have met. Twenty years of executive coaching. Thirty years and twelve books. And a career in technology that started in 1982.

I went in expecting to defend the AI. That is usually where these conversations start, and I have had enough of them to know the shape.

Not one of the three took the bait. They spent the hour somewhere else entirely, and by the end of the third call it was clear they had all walked to the same place from three completely different directions.

Availability was never the problem

The first coaches senior leaders inside global manufacturing and pharma, on a monthly retainer. He was directing theatre and coaching actors for twelve years before that.

His retainer already includes unlimited access between sessions. Nobody uses it. The more senior the client, the less time they have and the less they want to talk to their coach every week. That is precisely why he can charge what he charges.

So the thing we usually lead with, more support between the sessions, solves a problem his clients do not have.

He went somewhere better without being asked. First to what happens when the six months end, because people extending at a lower intensity is revenue that does not exist in his practice today. Then to something we had not considered at all. He started rehearsing the sales conversation out loud. Not a generic AI coach. A version of me, built from what you and I actually talk about, available from month three.

He was not asking whether the technology works. He was working out what he could promise a buyer.

Coaching ending is a convention, not a law

The second has been around technology and professional services since 1982, coaches leaders inside tech, and has spent the last two years as a fellow at a coaching faculty. He has also built his own version of what we build, out of about a dozen prompts and a lot of curiosity.

When the conversation reached the point where coaching is supposed to stop, because the profession is careful about dependency, he pushed straight back. Steph Curry has a coach. LeBron James has a coach. People operating at the highest level keep an external voice for life, and he does not accept that the ceiling on the relationship is anything more than the way coaching currently gets sold.

Then he named the thing a general model cannot fake. Of all the competencies he teaches, the one carrying the most weight in practice is organisational acumen, which is really just domain knowledge. Understanding the world his client works in, then their company, then their own agenda. You get generic understanding from a chatbot. You do not get that.

A body of work is a record of who you used to be

The third has thirty years, twelve books, around twenty five courses, and a community of change makers she has built from the ground up.

She has had access to a tool for building her own digital twin for eighteen months. She has never built it. Not for lack of time.

The tool wants the body of work. Hers spans thirty years, so feeding it in teaches the machine to contradict itself. When she looks at what comes back, she recognises herself from a decade ago rather than the way she works now. She works with what is emerging, and what is alive for her this week may have moved by next week. A snapshot of everything she has ever written is the wrong shape for her.

Training it on her sessions instead had not occurred to her, because nobody had offered it.

She also got to our premise before I explained it. She has no interest in a model where the client either works with the AI or works with her. She sees it integrated, she has already integrated AI into her own practice, and she said she has not yet found anyone offering the live element and the AI together.

What all three were actually arguing about

Continuity.

None of them was asking whether an AI can coach. All three were asking whether the relationship has to end when the engagement does.

One came at it commercially, one through the competency frame and two basketball players, one through her refusal of an either or. That is the question underneath CoachNova, and it was reassuring to have three people arrive at it without any help from me.

And each of them set a condition

This is the part I keep coming back to.

It has to sound like me. The first coach's instrument is the voice. He spent twelve years as an acting coach and he was clear that a version of him a client only reads is a smaller thing than a version they can hear. Today you speak to your AI and it writes back, because that is what people reach for first. His argument has moved a spoken version up our list, and he will hear about it from Nick before anyone else.

It must not slide into flattery. The second coach's endorsement came with a condition attached. If it holds the coaching stance and does not drift into agreement and flattery, he called it extremely valuable. If it drifts, it is worth nothing. He is right, and it is the failure mode we test hardest, because a general model is trained to be agreeable and coaching is often the question that does not go along with things.

It has to be who I am now. The third coach's point is the sharpest thing anyone has said to us about how these are built. Your AI starts from your sessions, which is the only record of how you coach today. Your books and frameworks are something you add on top when you want to, not the foundation the whole thing rests on. Your practice moved. The record of it should move too.

Three conditions, from three people who owe us nothing.

None of them is an objection. Together they are a specification, and a better one than we wrote ourselves.