■ MODERATE RISK ■ Personal Services
Not replaced, but repriced. Program design — historically a big part of what trainers charged for — is now effectively free, so the job is consolidating around coaching, correction, and accountability.
“AI builds workout plans. But yelling 'one more rep' requires human energy and eye contact.”
Our AI replacement risk score — how we score jobs
A trainer's session is three things happening at once: prescribing load and progression, watching movement and correcting it in real time, and managing a person's motivation and adherence. The last of those is why clients actually pay. Anyone can find a training program; almost nobody executes one consistently alone. Trainers also handle the messier context — a client with a cranky shoulder, a week of bad sleep, a wedding in ten weeks — and adjust on the fly, which is closer to case management than to spreadsheet building.
Programming is the exposed piece and it has been for a while. AI apps now generate periodised plans, auto-regulate loads from logged performance, and adapt to missed sessions, and they do it competently for general population goals. Wearables track heart rate variability, sleep, and recovery, and computer-vision form-checking apps can flag a rounding spine or a knee collapsing on a squat from a phone camera. Nutrition tracking and macro planning, once a paid add-on, is now an app feature. For a client who is self-motivated and injury-free, an app plus a gym membership is a legitimately good substitute for a mid-tier trainer.
What survives is physical presence and human accountability. Hands-on cueing, spotting under heavy load, rehab-adjacent work with real pathology, and the simple fact of someone expecting you at seven on a Tuesday — none of that comes from software. Population trends help too: ageing clients, post-injury referrals, and clinical exercise work are growing and require genuine expertise. Our risk score of 35 marks a field where the commodity tier collapses and the specialist tier does fine.
Automatability: our editorial assessment of current and near-term AI capability
Transformed rather than eliminated over the next decade or so. Generic online coaching and template-plan selling are already being undercut by apps that do the same thing for a monthly fee, and that erosion continues hard through 2030. In-person coaching, small-group training, and clinically adjacent work hold value well past 2040 because they sell presence and expertise. The likely outcome is fewer trainers earning more each, and very few earning anything from programming alone.
For a healthy, motivated, experienced lifter, largely yes — plan generation and progression are solved problems and apps do them cheaply. For beginners, anyone with an injury history, or anyone whose actual problem is turning up, it is a poor substitute. Most people who quit training do not quit because their program was wrong.
Yes, if you build a specialism and a client base rather than selling generic sessions. The commoditised end — cookie-cutter online plans, entry-level gym floor work — faces genuine pressure from apps and from oversupply of newly certified trainers. Specialists in rehab-adjacent, older-adult, and athletic populations are in demand and command real rates.
It helps and it is improving, particularly for obvious faults in well-lit, single-plane movements. But camera-based analysis misses intent, pain, breathing, and the small tactile corrections a coach makes with a hand on a hip. It also cannot decide, mid-set, that today is not the day for that weight. Useful supplement, weak replacement.
Reprice honestly: charge for coaching time and outcomes, not for the PDF. Adopt AI tools for programming and client admin so you can serve more people well. Deepen one specialism to the point where you get referrals from clinicians. And invest in the interpersonal side, because adherence coaching is the durable product you are actually selling.