How business coaches should use AI with clients

Most of the conversations I have with business coaches about AI start with some version of the same worry: is this going to replace me? It's a reasonable question. AI can now ask a decent coaching question, summarise a long conversation and suggest next steps, all in a few seconds and for very little money.

My answer is no, not for coaches who work with founders and leadership teams. But I also think coaches who ignore AI will find themselves at a growing disadvantage, because a lot of what fills a coach's week isn't coaching at all. It's preparing, chasing, writing up and finding out what happened. That part, AI is very good at.

This piece is about where I think the line should sit. It's the same line we built Clearahead around, which I've written about in The machine asks, people decide: AI does the asking, the gathering and the remembering; people do the deciding.

Where AI helps a business coach

Preparing for sessions

The ten minutes before a session are some of the most valuable a coach spends, and the hardest to protect when you have a full diary. AI can read what a client's team has done since the last session and tell you what moved, what's been quiet and what's waiting on a decision, so you walk in knowing where the real conversation is. There's a routine for this in How to prepare for a business coaching session.

Keeping the client's plan current

The single biggest reason coaching results fade is that the plan agreed in the session stops being used. AI changes the economics of keeping it current. Instead of someone updating a spreadsheet, people can say what happened in a sentence and AI works out which part of the plan it belongs to, updates it and keeps the facts as evidence. That's the difference between a plan that's current every day and one that's current the day before your session.

Writing things up

Session notes, decisions, a weekly summary for the client's team. These matter, and they take time. AI can draft them in seconds from what was said, and you or the client can correct them.

Asking structured questions

When a client is setting a goal or working out what's in the way, there's a set of questions that almost always helps. Why does this matter? Why do you believe it's possible? How will you know it's done? What's really stopping it, and why? AI is good at asking these consistently, which frees you to listen for the answer that matters.

Between-session support

A founder who's stuck on a Tuesday afternoon doesn't always need their coach. Sometimes they need someone to ask them a couple of good questions and help them think. AI can do that at any hour, and it can leave a record so you can see what came up.

What to keep for yourself

The hard question

AI asks good questions, but it asks them politely and evenly. The question a founder needs to hear, the one that lands because of who's asking it and when, is still yours. So is knowing when to stay quiet.

Reading the room

When a leadership team says it agrees but the head of sales has gone quiet, a good coach notices. AI doesn't see who looked at whom, or who said yes while meaning no.

Which conversation matters

AI can tell you that five things are stuck. Deciding which one the team needs to talk about today, and how to start that conversation, is judgement built on years of working with people. It's what clients are really paying for.

The relationship

Clients stay with coaches they trust. Trust comes from being understood by a person who has seen them at their best and worst and is still on their side. That doesn't automate.

What AI should never do in a client's business

• Make the decision. AI can propose the next moves or flag that the order of priorities looks wrong. The client's team decides. Its view is information, not a gate.
• Judge individual people. Using AI to score a client's staff, or to report on them to the founder, destroys the trust that makes the work possible.
• Watch people without their knowledge. If AI is reading what a client's team says, they should know, and so should you.
• Pretend to be you. If a client is talking to AI, they should know it's AI.

A note on client data

Before you put any client information into an AI tool, check four things: where the data is stored, whether it's used to train models, who can see it, and whether the client can export or delete it. Tools built for business use usually answer all four clearly. Consumer chat apps may not, and a founder's strategy, numbers and team issues are exactly the kind of thing you shouldn't paste somewhere you can't vouch for. Tell your clients what you use, and get their agreement. It's a good conversation to have anyway.

Why a general chat assistant isn't enough

Plenty of coaches use ChatGPT or Claude for their own thinking and preparation, and they're excellent for it. They're less suited to holding a client's plan. Each conversation starts afresh unless you set it up carefully, nothing is shared with the client's team, and there's no record of what moved and when. The plan ends up in your chat history rather than in the client's business. I've compared the two approaches in ChatGPT and Claude vs Clearahead.

What this means for coaching

If AI takes over the admin around coaching, two things change. First, a coach can work with more clients without the quality of each session dropping, because preparation and write-ups take minutes rather than hours. Second, sessions get better, because the time that used to go on catching up goes on the conversation that matters. The coaches who'll struggle aren't the ones AI replaces. They're the ones still spending the first half of every session finding out what happened.

How Clearahead draws the line

Clearahead is built on exactly this division of work. AJ, the AI inside it, asks the questions that sharpen a client's goal, helps the team find the constraints holding it back and suggests moves to clear the top one. It reads what people say in a sentence, files it as evidence on the right part of the plan and writes This week from AJ every Monday: the wins, what moved, what was learned and up to three things that need a decision.

What AJ doesn't do is decide. The client's team sets the goal, agrees the order of the constraints and chooses the moves. You, as the coach, read the same picture from one login across all your clients and bring the judgement. Clients own their data, which is encrypted and held in European data centres, and AJ uses OpenAI's API, which doesn't train on it. Read more on Meet AJ and Clearahead for coaches and advisers.

Questions coaches ask

How can business coaches use AI?

Mostly on the work around the coaching: preparing for sessions, summarising what a client’s team has done, keeping the client’s plan current between sessions, writing up decisions and drafting briefs. AI can also ask good structured questions as a client sets a goal or looks for what is holding it back. The judgement, the hard questions and the relationship should stay with the coach.

Will AI replace business coaches?

Unlikely for coaches who work with leadership teams. AI is good at gathering information, asking structured questions and writing things up, which takes over much of the admin around coaching. It is much weaker at reading a room, challenging a founder who doesn’t want to hear it and knowing which conversation a team needs. Those are the parts clients value most.

Is it safe to put client information into AI tools?

It depends on the tool. Before using any AI with client information, check where the data is stored, whether it is used to train models, who can see it and whether the client can export or delete it. Tools built for business use usually make these clear; consumer chat apps may not. Tell clients what you use and get their agreement.

Should coaches use ChatGPT or Claude with clients?

General assistants like ChatGPT and Claude are useful for a coach’s own preparation and thinking. They are less suited to keeping a client’s plan, because each conversation starts afresh, nothing is shared with the client’s team and there is no record of what moved. For that, a tool built to hold the plan and keep it current works better.

What should a business coach never hand to AI?

The decision. AI can propose, summarise and question, but the goal, the order of priorities and the calls on what to do should be made by the client’s team, with the coach’s help. AI also shouldn’t be used to judge individual people’s performance or to monitor a client’s team without their knowledge.