What AI Should — and Shouldn’t — Do in Executive Coaching

The boundary question

The most important question about AI in coaching is not what it can do. It’s where the boundary is between what AI supports and what it replaces.

Get that boundary right, and AI makes coaching better. Get it wrong, and you don’t have coaching anymore — you have a feedback system wearing a coaching costume.

What AI should do

AI is well suited to the reflective and analytical layer of coaching:

• Pattern recognition — surfacing habits in communication, question patterns, or recurring themes across sessions.
• Between-session reflection — providing structured prompts that help a leader reflect between conversations.
• Preparation support — helping a coach or leader prepare for a difficult conversation by mapping stakeholders and possible responses.
• Measurement — tracking whether the behaviors identified in coaching actually show up over time.

In each case, AI is a tool that extends the coach’s or leader’s capacity. The human still decides what it means and what to do.

What AI shouldn’t do

AI should not:

• Replace the relational judgment that determines whether an insight is useful or premature.
• Decide what a leader should do — that’s the leader’s work, and the coach’s role is to help them do it better.
• Become the relationship — the trust, the challenge, the holding of complexity that makes coaching work is relational, not informational.
• Score performance — coaching is development, not evaluation, and confusing the two changes everything.

The design principle

The principle is simple: AI should augment reflection, not replace relationship. If a tool makes the coach or leader more reflective without making the relationship less central, it’s doing its job. If it starts to stand in for the relationship, it has crossed the line.

That line is not a technology question. It’s a design question — and a human one.