Trainerize FitMetrics, Everfit AI and the Problem With Replacing Coaches

AI is going to change online coaching.

That isn't really up for debate anymore.

The bigger question is:

What parts of coaching should AI actually replace?

Because I think this is where a lot of coaching platforms are getting AI badly wrong.

The mistake isn't adding AI.

The mistake is building an AI coaching company when what you actually have is a coaching platform used by human coaches.

When your focus becomes:

How much AI can we add?

How much can we automate?

How many more clients can one coach manage?

…it becomes very easy to optimise the wrong things.

You stop asking whether the technology actually helps coaches do better work.

And eventually, in trying to improve the platform, you start undermining the people who make a living using it.

To me, that's the fundamental difference between what we're seeing from some platforms and what we're building at Coachflow.

AI should improve the coach.

It shouldn't replace the coach.

Why coaches are angry about Trainerize FitMetrics

In September 2026, ABC Fitness acquired FitMetrics and integrated the technology with ABC Trainerize.

FitMetrics brings AI into areas including client check-ins, messaging, nutrition analysis, compliance tracking and workout programming.

Some of that sounds great.

I'm not against AI.

In fact, we're building AI into Coachflow.

I don't think coaches should spend hours manually doing repetitive work that software could handle in seconds.

But it's the direction of the technology, and what we're choosing to optimise for, that concerns me.

Trainerize currently promotes FitMetrics as allowing:

One coach to manage 200+ clients instead of 50.

It compares that with hiring additional coaches and advertises an 89% reduction in overhead to service the same 200 clients.

It also talks about being able to:

Scale to hundreds of clients autonomously.

FitMetrics' own website goes further.

For lower-ticket services, it literally labels one model:

“AI as Coach.”

It promotes automated weekly check-ins, automatic macro adjustments, AI workout programming and a Full Autonomous Mode designed to run a low-ticket programme largely on autopilot.

And one of the examples being used to demonstrate the opportunity shows a coaching business moving from eight coaches servicing its client base to two while handling even more clients.

So I don't think coaches are imagining the direction this is moving in.

Reduced human input is being sold as one of the benefits.

“Working with a coach isn't just about what to do”

The Trainerize announcement has generated a lot of reaction from coaches online.

And one of the reactions that prompted this article made a really important point:

Working with a coach isn't just about knowing what to do.

Because knowing what to do has arguably never been less valuable.

Ask ChatGPT to write you a hypertrophy programme.

It will.

Ask it how much protein you should eat.

It'll tell you.

Ask it to build you a meal plan.

Done.

Ask it how to grow your glutes.

You'll have an answer in seconds.

Information isn't scarce anymore.

Your client isn't paying you because you know their macros

When somebody hires you as a coach, they aren't simply buying information.

They're buying into you.

Your experiences.

The way you think.

Your values.

Your approach.

The way you communicate.

And, importantly, somebody they trust to help them achieve an outcome they've been struggling to achieve alone.

If all they wanted was a workout programme, they could open ChatGPT for free and ask it to write one.

Any good coach worth their salt knows that 99% of people's problems aren't information problems.

They're implementation problems.

People generally know they should move more, eat better, train consistently and sleep enough.

The difficult part is getting a real human being, with a job, kids, stress, relationships, bad weeks, holidays, habits and emotions, to actually implement those things consistently.

That's what they're paying you for.

This is why the “200+ clients per coach” message worries me

I completely understand why 200 clients sounds attractive.

If you're a coach looking at the numbers, of course it does.

But this is where the fairy-tale version of scalability starts to fall apart.

Retention, great results and genuinely high-value touchpoints come from actually knowing the people you're coaching.

You know when your client's daughter's birthday is coming up.

You remember that their son just won their first football tournament.

You know they're going through a breakup.

You know somebody has recently lost a family member.

And sometimes the important thing isn't something they've explicitly written at all.

A great coach notices when the client who is normally relentlessly positive has suddenly become more negative.

They notice when the rockstar client who never misses anything starts drifting.

They pick up on changes in language, tone, behaviour and frequency before those changes become a cancellation.

That's coaching.

And that's the context I worry you lose when the objective becomes getting one person to oversee hundreds of people.

There is actually a useful parallel here in research around human relationships.

Robin Dunbar's work suggests there are cognitive and time constraints on the number of meaningful social relationships humans can maintain, with roughly 150 commonly cited as the average size of a person's broader active social network.

I'm obviously not claiming there's some scientifically proven 150-client coaching limit.

That's not what the research studied.

The point is much simpler:

Human relationships require attention.

And attention doesn't scale infinitely.

I feel like, as an industry, we've become addicted to big numbers.

More clients.

More revenue.

More scale.

But people rarely stop and think about what has to be attached to those numbers for them to actually mean something.

Everfit's AI Smart Response creates the same problem in a different way

Everfit's Smart Response is less extreme, but I think it raises the same underlying question.

When a client sends a message, Everfit can generate a suggested reply based on the conversation, saved responses and information available to the system.

You can select tones such as:

Formal. Casual. Excited.

And with automatic generation enabled, a draft can already be waiting when you open the conversation.

To be fair to Everfit, the coach still reviews and sends the message.

The AI isn't secretly having completely autonomous conversations with clients.

But I don't think that makes the question disappear.

The problem with AI writing the first draft

People will say:

“What's the problem if AI writes a really good response and the coach checks it before sending?”

Because checking somebody else's answer isn't the same cognitive process as creating your own answer.

If you've properly read the check-in, understood the problem, thought about the person and decided what they need, you've coached.

If AI gives you a polished answer first and your job becomes:

“Does this look good enough?”

You've changed the process.

And human beings are human beings.

Over time, people will skim.

They'll approve something that looks right at a glance.

They'll become reliant on the first answer they've been given.

And slowly, they stop learning their clients as deeply themselves.

At face value, you've saved five minutes.

Underneath that, you can start eroding the exact things that create retention, results, reputation and great coaching.

I'd also ask a pretty basic question:

If the part of coaching you want to remove is reading what your client said, thinking about it and responding… why did you get into coaching?

AI writing in “your voice” isn't the same as you coaching

This is another phrase we're going to hear constantly:

“It sounds like you.”

But training an AI to sound vaguely like you doesn't make it you.

It's a bit like wrapping a dog poo in a banana skin and claiming it's a banana.

The outside might look roughly right.

It's still not the thing somebody thought they were getting.

An AI can learn phrases you use.

It can learn sentence lengths.

It can analyse transcripts, previous messages and examples of your writing.

But that isn't the same as giving it your experiences, judgement, values, opinions and every bit of context that has shaped the way you coach.

Large language models are, in extremely simplified terms, incredibly sophisticated prediction engines.

Giving one a few hours of transcripts doesn't suddenly transfer your entire human experience into it.

And that's the bit that matters.

Because the decision is often the coaching.

Not the wording afterwards.

Even if we eventually reached some hypothetical future where an AI could display empathy almost indistinguishably from a human, there's another issue.

If somebody has chosen to pay a specific human coach because of that person's values, personality and approach, that's who they expect to work with.

They didn't buy a synthetic version of that person.

Lewis Potts and the difference between information and coaching

This is something I've spoken about before with Lewis Potts, creator of The Adherence Amplifier.

Lewis's work focuses heavily on adherence, retention and improving the actual skill of coaching.

When Lewis joined me on The Elite Coaching Blueprint Podcast, we talked about why clients fail to stick to plans and why better coaching often has very little to do with giving someone more information.

We talked about ownership.

Autonomy.

Behaviour.

Adherence.

And the psychology behind why somebody can know exactly what they're supposed to do and still not do it.

Listen to the episode with Lewis Potts on Spotify.

AI might become the separation filter between good coaches and bad coaches

Because a lot of bad coaching was never really coaching in the first place.

It was instruction. Just telling people "do this... do that..."

Here's your training plan.

Here are your calories.

Go and do it.

Then when the client doesn't do it:

“You need to be more consistent.”

Then when they fail again:

“You need to try harder.”

A great coach does something completely different.

They ask why.

Why didn't this work for this person?

What keeps happening?

What does this person's life actually look like?

What needs to change in the plan, environment, expectations or behaviour to make this achievable?

Bad coaches instruct clients.

Great coaches work with them.

And AI is incredibly capable of delivering instructions.

Which means those human coaching skills are going to become even more valuable.

But its also something that even if AI is able to replicate, that as human being we want connection with other humans - and that means our value is tied to the true coaching skill

NOT, being able to update someones macro's and calories

Personal brands make this even stranger

There's another contradiction here.

Simon Booth, founder of Strong Media, has spent years working with coaches around content, video and personal branding.

Recently his message has been heavily against Ai generated slop that most coaches are leaning into...

and honestly, I couldn't agree more

Social media, and coaching social media specifically has become more competitive than ever.

Especially where the barrier to entry for creating content is lower than ever, if your approach to stand out in an industry as a coach is to do what everyone else is doing, and outsource your thinking and personality to a bot...

I think Simon would agree, you're in a very dangerous position for your business.

Think about what we constantly tell coaches to do.

Build a personal brand.

Show your personality.

Create content.

Share your opinions.

Build trust.

Give people a reason to choose you rather than the thousands of other coaches they could hire.

Then somebody finally becomes a client…

And we start looking for ways to automate the relationship they bought into.

That's backwards.

If someone buys because of you:

They should get you.

Technology should make it easier for you to deliver that.

It shouldn't manufacture a synthetic version of you and call that scalability.

At Coachflow, we're taking a different approach

We're not anti-AI.

Far from it.

We're building AI into Coachflow because I think the opportunity AI gives coaches is massive.

But our philosophy is different.

We call it:

Human-Led AI

For me, Human-Led AI comes down to one simple question:

Who is the decision filter?

Imagine a client mentions in their check-in that an exercise isn't working for them.

AI could identify that request.

It could pull up the relevant training history.

It could surface it clearly to the coach.

Great.

Then I decide whether that exercise should change.

I'm still coaching.

The AI has simply removed friction around the decision.

What I don't want is the system identifying the issue, deciding what should replace the exercise and implementing that change simply because technically it can.

That's AI doing the coaching.

AI should speed us up.

It should organise information.

It should surface things we might otherwise miss.

It should help us understand context faster.

But the coach should remain the decision-maker.

AI should do the admin

Summarise information.

Organise data.

Calculate things.

Surface requests.

Identify patterns.

Bring important information to my attention.

Give me back the twenty minutes I'd otherwise spend clicking around six different screens trying to work out what happened this week.

Great.

AI should help the coach think

Show me context.

Highlight something unusual.

Show me that a behaviour has changed.

Bring up information I might otherwise have missed.

Help me make a better decision.

Also great.

The coach should still coach

Judgement.

Accountability.

Difficult conversations.

Understanding the individual.

Changing your approach based on what's in front of you.

Deciding what happens next.

That's the job.

AI-assisted coaching and AI coaching are not the same thing

I think about it the same way coaching changed when computers arrived.

A coach might once have written somebody's programme on paper.

Then they started typing it into a Word document.

Then they started delivering it through a coaching app.

The medium changed.

The coach didn't.

They were still analysing the client, making the decisions and doing the coaching.

Human-Led AI should be another version of that.

Better tools.

More information.

Faster workflows.

Less admin.

But the human still performs the job.

AI coaching crosses the line when the technology starts doing the thinking, deciding and communicating for you and the human becomes little more than the person pressing approve.

And coaches are actively opting into this

This is the bit that seems even crazier to me.

These aren't always features being forced on coaches.

Coaches can actively opt into them.

FitMetrics openly promotes AI doing check-ins, adjusting nutrition and, in some services, operating autonomously.

At that point, I think coaches have to ask themselves an uncomfortable question:

If your entire coaching service can be boiled down to an AI doing the work while you approve it, how much coaching were you actually doing in the first place?

Because you risk becoming the human face attached to an AI service.

The technology does the analysis.

The technology makes the recommendation.

The technology writes the response.

You read it and press a button.

You're no longer really using AI as a tool.

You're becoming the vessel through which the AI reaches the client.

And when platforms can boast about two coaches doing work that previously required eight, we should at least stop and ask what happened to the other six people's jobs.

Because there's an irony here.

Coaches using AI to remove themselves from coaching are also helping demonstrate exactly how easily their role can supposedly be removed.

That's not the version of coaching I want Coachflow to help create.

AI should increase your impact, not simply your client limit

Imagine AI saves you five hours every week.

You can look at that in two ways.

“Brilliant. How many more clients can I squeeze in?”

Or:

“Brilliant. How much better can I run this business without spending my entire life on my laptop?”

Maybe better systems mean you can take more clients.

There's nothing wrong with that.

Coachflow is software.

Of course I want technology to make coaching businesses more efficient and more profitable.

But capacity should be a consequence of better systems.

It shouldn't become the entire objective.

I don't want to build software whose end goal is helping coaches become increasingly absent from their own coaching service.

I want to remove the crap surrounding coaching so coaches have more time and information to actually coach.

Those are very different objectives.

AI will make great human coaches more valuable

I actually think AI is going to make great coaches better.

They'll have access to more information, faster.

They'll see trends that previously would've taken hours of manual analysis to identify.

They'll pick things up that might otherwise have gone unnoticed.

And they'll be able to do work that previously took thirty minutes in five.

That's what excites me about AI.

But there's also a bigger shift happening.

We're moving into a world where almost everything can be generated, optimised and made more efficient by AI.

And as that happens, I think genuine human input becomes more valuable.

Your creativity.

Your personality.

Your judgement.

Your experience.

Your empathy.

Your understanding of what it actually feels like to be another human being struggling with something.

Because if we automate all of that away, eventually everything becomes the same.

The same content.

The same messages.

The same recommendations.

The same AI slop wrapped in slightly different branding.

Human beings became successful because of an incredibly powerful combination of intelligence, experience, creativity, empathy and social connection.

AI can amplify those things.

I don't think we should willingly hand them over.

Not everything valuable is efficient

There's something else I think our obsession with optimisation misses.

Not everything valuable is efficient.

Sometimes the inefficient part is the entire point.

If you're having dinner with somebody you love, the objective isn't to consume the meal in the shortest possible time.

You sit there.

You talk.

You enjoy the food.

You probably waste time.

And that's exactly what makes the experience valuable.

Coaching can be like that.

The ten-minute voice note that technically could've been a three-line AI summary might be the thing that makes somebody feel understood.

Remembering something completely unrelated to their macros might be what strengthens the relationship.

Having a conversation that doesn't neatly fit inside an optimisation metric might be the thing that keeps someone going.

Efficiency is useful.

But efficiency should serve the outcome.

The outcome shouldn't be sacrificed in pursuit of efficiency.

The question every coaching platform should be asking

I don't think using AI makes somebody a bad coach.

And I don't think Trainerize or Everfit are wrong simply because they're experimenting with AI.

That would be a stupid argument for me to make while building AI into Coachflow.

The issue is how we use it.

The question shouldn't be:

How much of the coach can we remove?

It should be:

How much unnecessary work can we remove so the coach can have more impact?

That's Human-Led AI.

AI should make coaches faster, better informed and more capable.

But the human should still coach.

That's what we're building Coachflow around.

Coachflow is being built for coaches who still want to do the coaching. Come and help shape it.