Here’s a pattern a lot of coaches will recognise. A good month fills the diary. Every hour goes on sessions, so the posting, the follow-ups and the check-ins with past clients stop. A month or two later the diary empties, and there’s a scramble: a launch, a burst of posts, a run of discovery calls, and then the diary fills and the whole thing starts again.
So the business side gets done in bursts, and it’s hard to tell what’s working. If that sounds like your practice, this post is about the part AI can take off you, and the part it can’t.
Why the empty months follow the full ones
The work that brings clients in is the work you drop first when you’re busy. A session with a paying client is booked and in the diary. A follow-up with someone who enquired last week isn’t, and nobody will notice if it doesn’t happen until a month later, when there’s nobody waiting to start.
So the swings aren’t really about how good a coach you are, or even how good your marketing is. They come from the fact that your pipeline only runs when you remember to run it, and you remember least when you’re doing well.
That’s also why the bursts give so little back. In a quiet month you try everything at once (Instagram, a newsletter, a podcast guest spot, LinkedIn, a free challenge), each done in a rush, and most of it stops once the diary fills again.
What a system means when it’s just you
When people say a coaching business needs systems, it can sound like you need software, a CRM and a funnel. For a one-person practice it’s much smaller than that. A system is a short written list of the things that happen every week whether you’re full or not, plus a way of doing each one that doesn’t depend on you having a spare afternoon.
For most solo coaches the list is something like this:
- Every enquiry gets a reply within 2 working days, and a follow-up a week later if they haven’t booked.
- One piece of content goes out each week, in the one place your clients actually are.
- Every past client hears from you once a quarter, personally, not as a newsletter.
- Once a month you look at the numbers: enquiries, discovery calls, new clients, and where each one came from.
Your mentor or your community may give you a better list for your niche, and you should use theirs. Whatever’s on yours, write it down and keep it short. You decide what goes on it. AI’s job is to make each item cheap enough that you still do it in a full week.
Where AI fits, job by job
I’d set this up in a Claude or ChatGPT project, so your background, your programmes and a short note on how you write are loaded every time, and you’re not explaining yourself from scratch. The setup post for coaches walks through that, including the files to write first.
Intake
Whatever your intake form collects, AI can turn the answers into a one-page brief before the discovery call: what they said they want, what they’ve already tried, the one question worth opening with. You still run the call. You just don’t spend 20 minutes beforehand rereading the form.
Session notes
A voice note or a transcript straight after the session becomes a structured note and a short follow-up email in your voice. The prompts for session prep and post-session capture are in the earlier post, so I won’t repeat them here. Read the section below on client information before you paste anything in.
Follow-ups
This is the one I’d do first, because it’s where full months cost you the most. Someone enquired, had a discovery call, said they’d think about it, and then you got busy. For my own business I built a small agent onto the booking form that drafts the follow-up before I’ve even seen the lead. You don’t need anything that elaborate. A weekly list of everyone who hasn’t booked, and a prompt that drafts a short, personal note for each, gets you most of the way. You read each one and send it yourself. In the prompts below, style.md is the short note on how you write from the setup post, and past-posts.md is a handful of your own past posts.
Here's my list of people who enquired or had a discovery call but haven't booked yet:
<enquiries>
[first name, date, what they said they wanted, where we left it]
</enquiries>
For each person, draft a follow-up email of no more than 80 words. Use style.md for tone.
Rules:
- Refer to what they actually told me. Nothing generic.
- One question they can answer in a line.
- No discounts, no deadlines, no pressure.
- If my notes are too thin to write something personal, say so instead of guessing.
Content from your sessions
Some of your best content ideas are the questions clients ask you every week. At the end of each week, write down 3 or 4 of those questions in your own words, with no names and no details that could identify anyone. AI can turn one of them into a draft post or newsletter. That gives you one piece a week from work you’ve already done, instead of staring at a blank page in a quiet month.
These are questions clients asked me this week, written in my own words with nothing that identifies anyone:
<questions>
[3 or 4 questions]
</questions>
Pick the one most likely to help someone who isn't a client yet.
Draft a post of about 200 words that answers it the way I would in a session. Use style.md for tone and past-posts.md for examples.
Don't promise results. Don't invent a client story.
End with one question for the reader, not a sales line.
Check-ins with past clients
Past clients are often where referrals and returning clients come from, and they’re the easiest people to forget when the diary’s full. Once a quarter, a list of past clients and a line on what each was working on becomes a set of short personal notes for you to edit and send. It’s a small job, and it gets done instead of slipping.
Admin
Some of this doesn’t need AI at all. Your booking tool’s reminder emails will handle reminders better than any prompt, and your invoicing software can chase late payments on its own. Where AI helps is the writing around the admin: a polite reply to a reschedule request, a programme welcome email, terms written in plain English. Set the automatic parts up once, and use AI for the bits that need words.
Be careful with client information
If you’re a health, wellness or embodiment coach, your session notes may hold information about clients’ health, and UK GDPR treats that as special category data with stricter rules than ordinary personal data. So it really matters what you paste into a chat tool.
At the very least, keep names and health details out of a personal plan, turn off the setting that lets the provider train on your conversations, and use initials or a redacted summary when you want help with notes. If you find you need client information in an AI tool every week, look at a business plan that comes with a data processing agreement, and get proper advice for your situation. I’m not a lawyer, and this isn’t legal advice, but I’d want any coach I worked with to have thought about it before the first paste.
What AI won’t fix
AI can make a system cheaper to run. It can’t tell you who your practice is for, what your offer is, or what to charge. If the weekly list is wrong, AI will just help you do the wrong things more often. That part is strategy, and it’s what a good coaching mentor is for.
It also won’t stop you sounding like every other coach. If anything it’ll make that more likely, because a model that’s guessing produces the average coach. Whether a draft sounds like you is something only you can judge, and judging AI output is a skill you can learn, much like giving feedback to a new assistant.
Where to start
This week, write your list of 4 or 5 things that have to happen every week, full or not. Then pick one of them (I’d pick follow-ups) and set it up in a project with a short note on how you write. Run it for a month before you add a second.
If you want help with the judgement part, that’s what AI Fluency for Entrepreneurs is for. It’s 4 weeks with one live call a week, and it doesn’t install anything or need you to be technical. Every week has a solo route. In week 2 you bring a piece of your own AI work (a newsletter, a sales page or a programme outline) and it gets checked on the live call, with a short written note from me afterwards. Week 3 covers what to ask before AI goes anywhere near client information. There’s a free session on Thursday 26 November at 12:30 UK, and everyone who registers gets the replay.
If you run a practice with a team, associate coaches or a VA with regular delegated work, AI Systems for Entrepreneurs installs the system itself over 6 weeks. And if you’d rather work one to one than in a cohort, the AI Leverage Call is 90 minutes on your own setup, and the AI Brain Starter is where I install one job for you.
Frequently asked questions
How can a coach use AI without sounding like every other coach?
Give it your own material to work from, such as your past newsletters, the questions clients actually ask you and a short note on how you write, and check every draft before it goes out. AI starts sounding generic when it's guessing, so the fix is better input and a habit of checking, not a better prompt.
Is it safe for a health coach to put client notes into ChatGPT or Claude?
Be careful. Health information is special category data under UK GDPR. Keep names and health details out of a personal chat plan, turn off the setting that lets the provider train on your chats, and use initials or a redacted summary. If you need client data in an AI tool regularly, look at a business plan with a data processing agreement and take proper advice.
What should a solo coach automate first with AI?
Follow-ups on enquiries that didn't book. It's usually the first job to slip in a full month, and an AI draft you check and send is quick to do.
Join the Drift Digest
One short email in any week I publish, with the highlights from each new post, plus new AI tells as models change and a quarterly nudge to re-measure your voice. No spam, unsubscribe anytime.
Something went wrong. Please try again, or email web@ctoondemand.co.uk.
