The transcript of Thursday’s session reaches your inbox before you have written a single note, with three themes pulled out and a follow-up email drafted. That is where AI coaching earns its place, in the notes, preparation and admin around a session, while the conversation itself stays with you.
The harder question is whether your client agreed to a recording that software would read, summarize and store. You will see what the research says about AI coaches, which tasks to hand over and what to tell a client first. The section on when AI is worth adding to your practice turns all of it into a decision.
TL;DR
AI coaching covers three different things sold under one name. Some tools help the coach with notes and preparation, some give a client prompts between sessions, and some chatbots stand in for a coach on one narrow task. In two ten-month trials with business students, a goal-setting chatbot helped people reach their goals about as well as human coaches did. Those trials never tested empathy, challenge or a client in difficulty. So let AI handle the transcripts, summaries, intake and messages around a session, and keep the live conversation and every judgment about a person for yourself. Get a client’s consent before anything is recorded and know where each recording is stored. Set up basic business automation before paying for AI, and judge any tool by the time it returns and a measure it did not produce.
What Does AI Coaching Mean Today?
AI coaching describes three different arrangements, and each one puts the software in a different seat with different limits. Separate them first, because the seat decides what the software can honestly do.
| What is sold | Who uses it | What it can honestly do |
|---|---|---|
| Tools for the coach | The coach | Transcribe and summarize sessions, condense intake and recall what a client said months ago |
| A companion between sessions | The client | Prompt reflection, track a habit and keep a goal in view while the coach is not there |
| An AI coach in place of a human | The client, often through an employer | Run one structured routine, such as setting goals and checking progress against them |
AI coaching sold to employers for managers and individual contributors usually sits in the second or third row. Each person gets prompts and check-ins built from their own goals and answers, which is the usual sense of personalized AI coaching.
A coach buying software is almost always buying the first row. The second row is a decision you make with a client. A tool they use alone between sessions still shapes the work you do together, so agree on it when you contract.
The third row claims that software can do a coach’s job. It is also the one row with careful research behind it, and that research is worth reading closely.
Does an AI Coach Work as Well as a Human One?
On one narrow measure, an AI coach did about as well as human coaches in a pair of controlled trials. The limits of those trials matter as much as the result.
Nicky Terblanche and three colleagues ran two trials of ten months each, published in PLOS ONE in 2022. In the first, 105 students worked with a human coach while 105 others had no coach. In the second, 134 students used a goal-setting chatbot and another 134 had no coach.
Both coached groups made more progress on their goals than the uncoached groups. By the end of the ten months, the chatbot group had matched the human-coached group, a result the authors themselves called surprising.
Read the design before you quote that result. Every participant was a student at a business school in the United Kingdom, and every score came from participants rating their own progress. The human and chatbot trials also ran about two years apart, with different students.
The chatbot did one job only. It helped people set goals, commit to actions and check in on their progress, while the human coaches were free to work on whatever their client brought.
The authors drew a careful conclusion. They suggest AI could replace coaches who work from a simple, model-based routine, while its lack of empathy and emotional intelligence keeps human coaches hard to replace for now. Nothing in either trial tested a client in distress, a hard challenge or a coach changing the plan mid-session.
Where AI Earns Its Place in a Coach’s Week
AI earns its place in the work around each session, where the input is messy language and nothing depends on reading a person in real time. Before you buy anything, track one ordinary week and note where the minutes go between sessions.
These tasks are the usual candidates, because each produces something you can check against what you know:
- Transcripts and summaries: a recording becomes text, and the text becomes a short list of themes to read before the next session.
- Intake synthesis: a new client’s form, questionnaire and pre-work arrive as one page instead of a stack of separate files.
- Recall across sessions: ask what a client said about their manager months ago and see every place they said it.
- Draft follow-ups: you edit a first draft instead of writing from a blank page.
- Scheduling and reminders: routine messages that need no judgment about the client.
The last item barely needs AI, because booking pages, calendars and payment links have handled it for years. Price the workflow automation tools built for that job before paying for AI to do the same work.
One caution applies to every summary. A summary records what was said, not what happened, so you still have to remember the pause before an answer and the sentence a client started twice.
Treat a generated summary as an index into a session you sat through, never a substitute for remembering it. Other AI strategies in coaching go further, from drafting client narratives to exploring alternative possibilities, and each needs the same check against what you saw in the room.
What Should Stay With the Coach?
Anything that depends on you being present with a person, noticing what happens and responding in real time, stays with you. If a task needs you in the room, listening and deciding as the session unfolds, it does not delegate.
- Contracting: what the engagement is for, what sits outside it and what you will both do if it stops working.
- Challenge: a hard question lands because of who asks it and what the relationship can carry.
- Silence: a pause is often where a client does their thinking, and software that prompts in the moment fills it.
- Judgments about a person: readiness, honesty and whether an engagement has run its course.
- Referring a client on: when a client’s needs move beyond coaching, the decision to suggest another professional is yours alone.
Two failures sit on either side of this line. A coach who prepares only from generated summaries arrives with a record of the client rather than a memory of them. A coach who refuses every tool keeps paying for it in evenings spent typing notes.
Further out from the session, running the business involves no judgment about a client at all. Bookings, invoices and reminders can run with little or no attention from you.
Write your own line down in one sentence before the next tool arrives. If you cannot say it aloud to a colleague, it is not yet clear enough to explain to a client.
What to Tell a Client Before Anything Is Recorded
Before any session is captured, a client should hear what will be kept, where it goes and who can read it. Consent asked for afterward is an apology, not consent.
A recording changes what a session is. The client spoke expecting their words to stay in your memory, and now they also sit in a file that other people could open.
1. Before the First Recording
Cover five points out loud before the first recording, then write them into the coaching agreement:
- What is captured, whether audio only or audio and a written transcript.
- What comes out of it, such as a summary, a list of themes or agreed actions.
- Who can open it, named by role, including anyone at a sponsoring organization.
- How long it is kept, and what happens to every copy when the engagement ends.
- Whether a person ever reads it, or whether the processing runs without anyone looking.
2. Where the Recording Lives
A general note-taking app is the wrong home for a coaching transcript, however convenient it feels. Client records belong in a system with access controls, a retention setting and a delete that removes every copy of the file.
Ask each vendor where the audio sits after processing, and get the answer in writing. A line saying the data is encrypted answers a different question, so ask for the retention period in days.
Where client records live is one of the first things to settle when choosing coaching software, before features such as scheduling or billing. A tool that cannot answer the storage question is not ready for client sessions, however good its summaries are.
3. When a Client Withdraws Consent
Clients can change their minds about being recorded, and planning for that protects the relationship. Agree in advance that a client can ask for recordings and summaries to be deleted at any point.
When a client asks, delete everything, confirm it in writing and carry on with notes you write yourself. Being able to do that on request also gives you an answer when a sponsor asks how client data is handled.
Automate the Business Before You Add AI
Automating the business side of a practice is a different project from putting AI into the coaching itself. It usually comes first, because it costs less and touches no client conversation.
Intake forms, scheduling, billing and reminders all have settled answers that need no AI. Automating a coaching business means setting those steps up once, so that routine work happens without you.
Putting AI into the coaching work is harder, because each use touches what a client shares with you. One question separates the two projects. Would a client care about the change, and would you mention it when you contract?
A new payment processor does not reach that bar. A recording of every session does, so it belongs in the agreement your client signs, while the payment processor belongs in your budget.
When Is AI Worth Adding to Your Practice?
Adding AI pays off once remembering across sessions has become a real problem and you can already say where every recording will live. Add it when all four of these hold:
- You have more active clients than you can hold clearly in mind from one session to the next.
- Your engagements run for months rather than weeks, so recall across sessions matters more than recall within one.
- You already write session notes, and they have become the task you always put off.
- You can name where recordings will be stored and everyone who can open them.
Hold off if any of these apply:
- Your client list is small enough that your own notes and memory still serve you well.
- You cannot yet answer a client who asks where a recording is stored and who can read it.
- Your work often touches material that needs another professional, where a stored transcript adds risk that no saved hour outweighs.
The two lists describe the same practice at different sizes, and the answer changes as it grows. Reread both every few months, especially after taking on new clients.
How Will You Know AI Is Helping?
Judge an AI tool on two things you can check yourself, the hours it returns and what your clients do differently, rather than on the progress scores it produces. Who does the scoring can change the result.
A 2023 review of workplace coaching studies found positive outcomes whoever did the rating, whether managers or the clients themselves. The one study that used ratings from the client’s own team found a smaller effect, too uncertain to count as a finding.
A progress score that a tool builds from your own session notes is one source, not two. Keep at least one measure the tool did not produce, such as the client rating the engagement in their own words.
Then check the time honestly. If a tool saves twenty minutes a week but takes an hour of setup and checking, it is not paying for itself yet.
Proving progress to a sponsor is a separate job with its own rules. When someone else is paying, the evidence a sponsor accepts has to come from records in coaching tracking tools, which an AI tool’s own scoring cannot replace.
Conclusion
AI coaching is most useful when it gives you time back around the session and stays out of the conversation. The trials support a narrow goal-setting chatbot on one self-rated measure, with students, and they say nothing yet about the parts of coaching that depend on noticing a person.
Decide where your line sits, write it into your agreement and get consent before anything is recorded. Then keep the record somewhere built for client work. Simply.Coach keeps each client’s intake, goals and session records in one workspace, next to anything a tool drafts for you.
Key Takeaways
- Name the arrangement. A coach’s tool, a client’s companion and a stand-in coach have different limits, so say which one you mean.
- Quote the trials carefully. The chatbot result came from students scoring their own goals on one narrow task.
- Delegate the paperwork. Transcripts, intake and first drafts carry no judgment about the client.
- Keep the live work. Contracting, challenge, silence and referral all need someone present.
- Put consent in writing. Cover the five points before the first recording, and agree how deletion works.
- Automate before AI. Booking and payments need no AI and never touch a client conversation.
- Keep an outside measure. A score built from your own notes cannot confirm itself.
If a client’s history is spread across transcripts, notes and email threads, AI only makes the pile grow faster. Simply.Coach brings intake forms, goals, stakeholder feedback and session records together on one leadership coaching platform, so you prepare from the whole engagement.
Frequently Asked Questions (FAQs)
What does an AI coach do?
It depends on which kind you mean. A tool for coaches transcribes and summarizes sessions and recalls past conversations. A companion gives a client prompts between sessions. A stand-in chatbot runs one structured routine, such as setting goals and checking progress, without a human coach.
Can ChatGPT be my life coach?
A general-purpose assistant can help you set a goal, reflect and plan next steps. It cannot notice what you avoid saying or decide when you need a hard challenge, so it works better as support between sessions than as a replacement for a coach.
Is AI coaching as effective as human coaching?
On one narrow measure, it can be. In two ten-month trials, a goal-setting chatbot helped business students reach their goals about as well as human coaches did. Every score was self-rated and the trials did not test empathy or challenge, so the result does not cover everything a coach does.
Will AI replace human coaches?
The researchers behind the main AI coaching trials suggest it could replace coaches who follow a simple, model-based routine. They also argue that its lack of empathy and emotional intelligence keeps human coaches hard to replace for now, especially in work that depends on the relationship.
Can AI coaching help managers and leaders?
It can give managers regular prompts and check-ins tied to their own goals, and it can reach a whole team at once. The strongest trial so far studied students rather than managers, so judge it by what changes in how people lead, not by how often they use it.
Do I need a client's consent to record a coaching session?
Yes, and ask before the recording exists. Tell the client what is captured, what comes out of it, who can open it, how long it is kept and whether a person reads it, then put that in the coaching agreement. Check the privacy rules where you and the client live.
What should AI never do in a coaching session?
AI should not handle contracting, challenge, silence, judgments about a person or the decision to refer a client to another professional. Each of these needs someone present in real time who notices what is happening and responds to it.