Finding your approach

What AI-augmented coaching means

I mean using AI to support the work of a human coach and the learning of their client. I do not mean using AI to do your coaching.

That can take different forms. A coach might explore an uncertainty after a session, prepare by examining patterns across appropriately handled notes, or rehearse a conversation using a fictional scenario. Some coaches also explore tools a client can use between sessions.

Those are different activities, with different implications. Using AI to examine your own coaching is a different proposition from giving a client an assistant to converse with independently.

Be specific about the activity you are considering. “I want to use AI in my coaching” becomes easier to work with when you can explain what you want it to contribute, for whom, and at which point in the work.

Starting with your own practice

Start with a question about your own practice.

Perhaps you sometimes move into advice too quickly. Perhaps you want to explore different ways of responding when a client seems stuck. Perhaps you want to revisit a familiar framework and examine what it draws your attention towards.

Choose one of those questions and explore it using a fictional coaching exchange or scenario. Form your own view first, then ask the AI to help you examine it. This lets you practise working with the tool before introducing client information.

A useful first experiment has a clear purpose and an end point. You should be able to say afterwards what it helped you notice, where you disagreed, and whether you want to try it again.

Choosing how much AI to use

Learning about AI can help you decide where to use it and where to leave it out. A coach might use it extensively for research, occasionally for reflection and never during a client session.

What matters is whether you can explain those choices. What have you tried, what did you learn, and what would lead you to reconsider?

What your coaching experience contributes

Your experience helps you assess whether an interpretation fits the situation.

A model might identify a question as leading. You can consider what preceded it, how the client responded, what you had contracted to do and whether a more directive intervention was appropriate at that moment.

Experience does not make us immune to accepting a persuasive answer. It gives us more to bring into the examination, provided we actually use it.

My suggestion is to ask three questions: what is the evidence, what else might explain this, and what do I know from the relationship that is missing from the text? The value develops through that conversation.

(Before using client material, see AI, confidentiality and your clients’ trust.)