For a couple of years, AI in fitness mostly meant chatbots in consumer apps. This year the center of gravity moved. Technogym unveiled its AI Ecosystem for operators, promising personalized programs that follow the member across equipment. EGYM keeps extending its Genius platform through connected strength lines, and in July went as far as deploying agentic AI in its own financing operations. Whoop, meanwhile, has pushed AI advice and even on demand clinician access onto the wrists of millions of consumers who then walk into your club with expectations. The message is consistent: AI coaching is turning into infrastructure.
For an operator, the question is no longer whether this wave arrives. It is what to buy, what to skip, and how to keep your staff at the center of the experience while the vendors race each other.
What AI coaching does well right now
Used honestly, today's systems earn their keep in a few specific places. They adjust programs from data: if a member's sessions show declining load or flat heart rate response, a machine can flag it and adapt the next block faster than a trainer reviewing spreadsheets. They nudge adherence: reminders, streaks and micro goals tuned to individual patterns measurably lift session frequency. And they triage attention, which may matter most for your economics. No coach can watch four hundred members. A system that surfaces the fifteen whose attendance is sliding tells your team where an hour of human contact will actually move retention.
What it still does badly
The honest list is just as important. AI does not correct a deadlift in real time from an app, whatever the demo suggests. It struggles to motivate the ambivalent beginner whose problem is not information but confidence, the person who needs a name remembered and a small win noticed. And it produces generic plans when it lacks data, which leads to the least discussed truth of this cycle: AI coaching is only as good as the signal underneath it. A club with no effort tracking, no attendance granularity and no training history has nothing for the intelligence to work with. Buying AI before you have data means paying for a system with nothing to reason about.
The staffing question your trainers are already asking
Every vendor deck says AI will not replace trainers. That is roughly true, but it does change the job. The trainer of the next few years spends less time writing generic programs and more time on exceptions and relationships: the member flagged as at risk, the plateau that needs a human eye, the conversation that turns a cancellation into a restart. Clubs that prepare staff for that shift will get compounding value from these tools. Clubs that deploy AI as a cost cut will discover that members can get generic digital coaching anywhere, including free, and that the only thing a club can uniquely sell is people plus data.
A ninety day pilot that costs almost nothing
You do not need a platform migration to find out whether AI earns its keep in your club. Pick the single use case with the clearest money attached: members at risk of cancelling. Establish a baseline month where your team works exactly as it does today. Then, for the next two months, have the system produce a short weekly brief: who is sliding, why it thinks so, and one suggested action per member. Staff execute, and you count saves against the baseline. The pilot forces the honest questions into the open. If the system cannot explain why it flagged someone, your staff will not trust it. If it cannot connect to your existing attendance and effort data, you have learned that its intelligence is generic. And if it works, you now have a number to weigh against the subscription price, which is a far better negotiating position than a demo ever produces.
How to evaluate what vendors are selling
A handful of questions cut through most demos. Where does the data come from, and does the system work with the wearables and sensors your members already use, or only with the vendor's hardware? Who owns the member's training history if you switch platforms in three years? What does it cost per active user rather than per member, since inactive members inflate every AI pitch? And what exactly does it hand your staff: raw dashboards, or prioritized actions a front desk person can execute on a Tuesday morning?
For transparency, this is the architecture UPTIVO chose for NATE, its coaching engine, which was built on the effort data the platform already collects in class and on the floor, and it is designed to brief trainers rather than bypass them. We think that is the defensible architecture for clubs, whoever you buy from: intelligence that runs on your data and strengthens your staff, not a black box that rents both back to you.
The wave is real, and this time it is aimed at operations, not app stores. Buy the parts that multiply your people rather than the ones that try to replace them.