The Most Common AI Mistakes Executive Coacheses Make

You input their CV, recent feedback, and a description of their role into ChatGPT, then walk into the session with a ready-made picture of their strengths and blind spots. This stops you from asking the questions that would reveal what they truly believe about themselves and what keeps them awake at night.

The fix

Go into the first session with only their name and role, and spend the first twenty minutes listening before you propose anything about who they are.

Tools like Humu and BetterUp AI produce scored profiles on emotional intelligence, decision-making style, and team impact. You start treating these scores as objective truth, missing the contradiction between what the algorithm says and what you hear when they talk about their relationships with direct reports.

The fix

When an AI score conflicts with what you observe in real conversation, trust what you observe and ask the leader directly why the gap might exist.

You copy suggested coaching questions from CoachHub AI or Claude into your session plan because the questions sound professional. You lose the specificity that comes from genuine curiosity about this particular person's situation.

The fix

Write your three opening questions by hand before the session, based on what surprised or troubled you in the intake conversation.

You ask an AI tool to summarise feedback from a 360 review, and you read the summary instead of the actual comments from their team. You miss the tone, the particular phrasing, and the clues about what relationships need real attention.

The fix

Read at least half of the raw 360 comments yourself, underlining phrases that surprise you, before you discuss results with the leader.

You generate a coaching agreement template from ChatGPT and send it to the leader, missing the conversation about what they actually need from this work and what you both expect to happen. The contract becomes a formality rather than a shared commitment.

The fix

Have a phone call or video meeting to discuss the coaching scope, your approach, and their specific goals before you send any document.

When a leader describes a conflict with their board or a moment of self-doubt, you feel a pull to ask ChatGPT or Claude how to respond. The moment you do, you signal that an algorithm's suggestion matters more than your own presence and judgment in the room.

The fix

Promise yourself you will sit in silence for ten seconds before you speak, and speak from what you actually think, not from what you might look up later.

A leader describes a setback, and you offer a reframing that you actually prepared using Claude or ChatGPT earlier that week. You present it as insight you have arrived at through your coaching work, damaging trust if they ever find out the source.

The fix

If you use AI to generate reframes, try them out internally first and only name them in session if you have made them your own through real dialogue.

Your AI tool flagged decision-making speed as a development area, so when the leader mentions conflict with a peer, you keep steering back to how quickly they decide. You stop responding to what they actually came to work on.

The fix

When the leader raises a concern, stay with it for at least five minutes before you introduce any agenda of your own.

You are coaching someone through a difficult realisation about their impact on their team, but instead of sitting with the weight of that moment, you are mentally preparing questions from a coaching prompt generator. Your own intuition about what to ask next atrophies.

The fix

Turn off your phone and put your coaching notes away for the last ten minutes of each session, and trust what you know how to ask.

BetterUp or Humu scores someone low on resilience or high on risk aversion, and the leader feels it is wrong. You treat the AI judgment as the more objective view instead of recognising that the leader's own self-knowledge might be more accurate than an algorithm trained on patterns.

The fix

When a leader disagrees with an AI assessment, say 'I trust what you know about yourself here' and ask what the score might have missed.

You record the session and ask ChatGPT to summarise it, creating notes that sound polished but miss the small moments where a leader's voice changed, where they became more animated, where they paused longer than usual. These details are what coaching relies on.

The fix

Write your notes in the two hours after the session, including at least one specific phrase the leader used and one observation about their body language or tone.

Between sessions, you feed Claude the summary from the last conversation and ask what the leader should focus on next. You walk in with a preset agenda instead of asking them what has shifted since you last spoke.

The fix

Start every session by asking 'What has come up for you since we last spoke?' and listen for at least five minutes before you suggest a focus.

You use a coaching app that scores progress on a scale, or you ask an AI to evaluate whether they are hitting their goals. You outsource the judgment that should come from knowing this leader over time.

The fix

Write your own notes on their progress every quarter, using specific moments and conversations, and keep your own record separate from any AI scoring.

You ask ChatGPT to suggest books on executive presence or decision-making, and you recommend them to the leader without having read them. If the leader wants to discuss what the book means, you cannot speak from experience.

The fix

Only recommend a book or resource to a coaching client if you have read it in full and can cite a specific moment that changed how you think.

A leader asks whether their management style matches research on high-performing teams, and you generate an answer from Claude to sound authoritative. You lose the chance to acknowledge what you do not know and to model intellectual humility.

The fix

When you do not know something, say so, and offer to research it properly if it matters to the coaching work.