I’m not sold on the idea of AI health coaching apps on my phone

One of the rumors about Apple’s AI ambitions is that it’s considering a paid coaching subscription service. This would offer guidance not just for exercise, but for eating and sleeping, all based on the data recorded by your devices — presumably synced via Apple Health on your iPhone. I’ll call it Apple Coach for short, although there’s no official name yet.

I’m genuinely interested in seeing what Apple comes up with, assuming the service does reach the market. But call me skeptical — I doubt the value of any AI coaching service, based on my own fitness experience and what I’ve seen from existing apps.

The mixed record of AI coaches

Sometimes, one miss is too much

JuggernautAI / Pocket-lint

To the industry’s credit, there are a number of AI coaching apps out there that are genuinely well-liked, even at the competitive level. In the weightlifting world, for example, there’s an app called JuggernautAI. While it does have features for beginners, some of its users include powerlifters active in leagues like the USPA and USAPL. It can formulate an entire training block with a meet date in mind, including the competition versions of lifts, which are both stricter and tougher.

AI may be even more useful at the beginner level, where many people don’t know that they need a plan, much less how to build one. While I’ve criticized Apple Fitness+ for a variety of reasons, it does at least include basic plan building, with simple AI recommendations based on workouts you’ve done in the past.

The trouble comes with optimization. While apps like Juggernaut (or, for another example, Sheiko Gold) may be based on well-tested concepts, and enable more personalization than a stock spreadsheet, they also tend to rely on trial and error to “dial in” their programs. That can mean pushing yourself too hard on some sessions, or not hard enough on others, at least if you take their recommendations verbatim. Also, beginners and athletes alike may find they respond better or worse than other people in the same boat. That makes sense — some people respond better to high-volume training than high-intensity training, or vice versa.

Entry-level AI coaches will sometimes avoid errors by making more conservative recommendations, even omitting routines entirely, but that diminishes or outright kills their value beyond beginner fitness. Once you’re experienced, you usually don’t need to be told to push harder, get more sleep, or eat more calories — you need to know exactly what that looks like in your case.

All this makes it difficult to recommend paying for an AI coach, simply because there are proven training regimens out there that can work just fine with a little human intelligence.

Related

The Powerbeats Pro 2 are incredible fitness-focused iOS earbuds

The Powerbeats Pro 2 stand head and shoulders above other fitness-focused earbuds.

The philosophy of getting fit

It doesn’t have to be complicated

If you boil it down, fitness can actually be very simple. Want to gain muscle? Do strength training with proper form, and eat in a calorie surplus. Want to lose fat? Eat in a calorie deficit and do some form of regular exercise. To make progress as a runner, try to go a little further or faster each time. To make progress in strength, hit every muscle group at least twice per week, and gradually increase your sets, reps, or weight, depending on the exercise. No matter what, make sure your diet has a lot of protein, sufficient carbs, and low sugar.

The devil is in the details, of course, but there are free, ready-made training and nutrition programs out there that already factor in the things AI tries to adjust for, such as fatigue and weight percentages. With time, it’s not hard to gauge these things yourself, and even substitute your own exercises and foods when something is unpleasant or getting poor results. My own fitness routine is highly customized at this point — depending on which one, AI would probably do more harm than good, at least until it adapted to my idiosyncrasies.

It’s also important to remember that there are a lot of useful apps and accessories that don’t depend on AI, at least beyond the vanilla kind of AI we’ve had for decades. I track my workouts with an Apple Watch Ultra 2 and the Fitness app on my iPhone, combining that data with Cronometer for calorie tracking and a Withings scale for body composition. Because I understand the principles behind fitness, including good technique and programming, I don’t really need anything more than that to produce results. The data I’m dealing with isn’t perfect, but all it takes is following trends to make the occasional tweak.

Remember that many of the world’s elite athletes never had the opportunity to use AI coaching. Usain Bolt retired from sprinting before ChatGPT launched in 2022, and Arnold Schwarzenegger won seven Mr. Olympia competitions before anyone owned a cellphone.

Related

Does smart tech really add that much to your fitness regimen?

The full answer varies from person to person, but here’s my gym-buff take.

Could an ‘Apple Coach’ get it right?

Walking the tightrope

Pocket-lint

I think Apple will do reasonably well, but mostly because it tends not to set the bar very high. Yes, the iPhone and Apple Watch do nudge you to close your Activity rings each day — but unless you adjust them, even a quick run or walk may do the trick, and most fitness badges are easy to earn if you’re already active. The company’s goal is to keep you healthy, not make you an athlete. Fitness+ workouts can potentially be challenging, but there’s nothing geared toward people who want to run a three-hour marathon or deadlift 500 pounds.

The wild card is Apple’s AI team. Apple Intelligence has received a lukewarm reception so far, and some of its key features have been delayed, like contextual awareness for Siri. Even assuming the Apple Coach team is firing on all cylinders, the company might still be forced to pull some people off that project to deliver on promises it made in 2024.

Who knows — I’d like it if Apple proved my skepticism is unfounded. Hypothetically, we could learn more at WWDC 2025 this June.

Related

iPad Air 7th gen vs iPad 11th gen: How the two Apple tablets stack up

You might not need as much iPad as you think you do.