UX Review of Train Fitness — the best Automated Weightlifting Tracking app that uses AI & Machine Learning

Train picks up where Apple left off

Apple doesn’t seem to care about traditional weight lifting as an exercise to be tracked on the Apple Watch. For years many have patiently awaited an Apple watchOS update that adds tracking for lifting weights in the gym. Yet year after year, we see amazing and magical new features to track running, yoga, and every other type of exercise under the sun. Sure some of the training in Fitness+ use dumbbells, but that isn’t really traditional weight or power lifting.

Enter Train (Fitness)

How does the app work?

The first time I opened Train on my Apple Watch I knew I was going to enjoy the user experience. Using a single button to get started with no configuration blew my mind in comparison to previous apps in this space. Also, using the 💪emoji instead of loading a custom graphic? Brilliant.

It doesn’t get more simple than that

Pick a weight and start moving!

A dial to select a weight to start working out
Using native UI elements is smart
A simple interface that shows a weight and the text “Start Moving” to start a workout
No caption needed :)

Interface Anatomy

Again, a very simple screen (seeing a trend here?).

An apple watch screenshot showing the interface for submitted a tracked exercise with name, reps, pounds, edit, save, and delete
  1. Easily change reps: if the app didn’t correctly count reps you can adjust (BUT you should consider: perhaps your form wasn’t correct and you shouldn’t count those reps)
  2. Easily change weight: changing weight has a convenience interaction; you can “accelerate” the pounds increments if you hold down the + or - buttons for a longer time
  3. When the app identifies the incorrect exercise you can click edit and select the proper exercise (I talk more on False Positives below)

Supersets

Relatively recently, superset grouping and tracking was added to the app. It auto-groups the exercises in the superset nicely, and allows modification with a tap of the exercise.

An apple watch screenshot showing a superset of two exercises, squats and military press

Manual Exercise Entry

For certain exercises (like many legs exercises) that are hard to track from a watch, there’s a manual exercise entry option. You’ll get a dismissible note that these exercises are slated to be trackable soon.

An apple watch screenshot showing a tip for a manually recorded exercise

Finishing & Saving a Workout

When you’re done, swipe and complete you workout to save it to your phone. You’ll get an iOS notification and you’ll be greeted with your latest workout and a very nice summary with beautifully designed stats that are very sharable.

An apple watch screenshot showing a working save confirmation
An iPhone screenshot of the recorded workout routine with auto-generated title, summary, stats and muscle group highlights
Somehow showing the total weight makes me feel like a super hero

Activity, Stats, and Leaderboards

Train offers comprehensive stats in their iPhone app. Some useful stats and metrics like total workouts, calories, reps, sets, hours, and my favorite: TVL (Total Volume Lifted).

Screenshots of the stats inside Train, including streaks, monthly view, muscle group focus and more
Look Ma, I didn’t skip legs!

Shortcomings & Usability Issues

The app isn’t perfect. The good news is that it’s constantly evolving and improving, with obvious user research, feedback and community involvement.

Analysing Timeout

Analysing (“Analyzing” for us folks in the United States) always takes JUST too long before timeout. I have to re-wake the watch. Every. Single. Time.

An apple watch screenshot showing the train app analyzing an exerice

Last Weight Defaults

Defaulting to the last weight VS the last weight for a specific exercise

An apple watch screenshot showing the same weight default applied to back to back exercises
Use the last weight for Military Press, not for the last exercise (Front Squat)

Where’s the Learning in Train’s Machine Learning?

The main benefit of Machine Learning (ML) and AI is that the experience can improve and adjust over time as the models are trained. It’s a bit disappointing for using an app built on ML/AI that doesn’t feel like it’s learning in certain areas. When using Train, I have to constantly change the weight of an exercise which I’ve done a thousand times. I’d expect the app to make a recommendation based off all my previous entries since I rarely use a different weight.

False Positives

One of the most irritating current quirks is when the app constantly identifies a common exercise as something radically different.

An image of bench press and incline bench press not being equal
These aren’t the same thing
  • Front Squats vs Smith Machine Squats
  • Diamond Push-ups vs In-And-Out Abs

False Negatives

Interestingly I’ve gotten many more false negatives recently where the app doesn’t identify or track a recent set at all. As an app matures I’d expect less and less false negatives, but perhaps they’re over-compensating and over-tuning for some other bugs and features.

Training & Testing New Exercises

I was surprised to get an invite to install an additional app to help track and train new exercises. This is awesome! It’s an incredible community-building initiative and clever company strategy to crowdsource the training of their models for new exercises. Kudos to them for this inclusivity of their users in this process.

Train Fitness Exercise Demo App

When you join the progress you’ll get access to an additional iPhone/Watch app to record yourself testing new exercises. This is great if you have a home or private gym, otherwise it could be awkward…

A screenshot of the Train demo app showing new leg exercises like calf raises and smith machine squats
The Train Demo App
Train Demo Weekly Leaderboard and rewards program named “Train & Earn”

UX Rating: 4.5 out of 5

4.5 out of 5 star UX review

There’s small usability issues and room for improvement, but Train Fitness is by far the best auto-tracking weight-lifting workout app on the market.

  • ✅ Super easy learning curve
  • ✅ Overall very intuitive and accessible experience
  • ✅ Allows for easy editing and manual entry where needed
  • ✅ Impressively accurate results and stats, best in class
  • ✅ Active community and nice early access demo program
  • 🫤 Mildly irritating False Positives and False Negatives
4.7 rating on the apple app store

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Kyle Ledbetter

Kyle Ledbetter

1K Followers

VP of Design at Credo AI — Former UX leader for Teradata, eBay, MicroStrategy and Sears Holdings.