← Back to blog
Apps & Tools· 5 min read

AI Form Correction Apps: What They Fix and How to Choose

By Mac Burke, qualified sports nutritionist


The short answer

An AI form correction app uses your phone camera to track your body and give cues when a rep drifts from a target, such as missing depth or knees caving. The good ones are clear about which exercises and camera angles they support, process video privately, and pair feedback with a way to reach a human coach for judgement calls.

An AI form correction app is a camera, a pose model and a set of rules. It watches you train, compares each rep to a target, and tells you when you drift. That is genuinely useful for the common faults, like shallow squats or half push-ups. It is not a replacement for judgement. Knowing which jobs these apps do well is the key to choosing one.

What does AI form correction actually mean?

Most apps follow the same steps. A pose model marks body points in each frame. Google's ML Kit pose detection, for example, returns a 33-point skeleton and runs at roughly 30 frames per second on a Pixel 4, per its documentation. The app turns those points into angles and positions, then checks them against rules for each exercise. If your hip does not drop below your knee, or your knees move inside your feet, it gives a cue on screen or by voice.

So correction really means rule checking. The app is only as good as the rules it uses and the view it gets of your body.

That is why two apps can give different feedback on the same squat. One may count a rep at parallel, while another wants your hips well below your knees. Look for an app that tells you what its depth or lockout rule is, so you know what you are being judged against and can match it to your goal.

Which form faults can an app catch reliably?

  • Shallow reps. Missing depth on squats, lunges and push-ups is easy to see from the side.
  • Incomplete lockout. Arms not straightening at the top of a press or push-up.
  • Range shrinking late in a set as you tire.
  • Knees caving inwards, if you film front-on with clear sight of your legs.
  • Large hip sag or pike in a push-up or plank, filmed side-on.

These faults involve big, visible movements. Schurr and colleagues (2017) found that plain side-on video measured trunk, hip, knee and ankle movement within 0.7-3.1 degrees of 3D lab capture. That is plenty of precision to judge depth and lockout.

Which faults are hard for an app to catch?

  • Small changes in lower back curve under a barbell.
  • Grip width, wrist angle and bar path on bench press, where the bench and bar hide joints.
  • Rotation and twisting, which a single camera sees poorly.
  • Exact joint angles. Asaeda and colleagues (2024) found MediaPipe Pose read knee angles about 19 degrees higher than a lab system during landings, though changes in angle tracked well.
  • Whether a position is right for your body, given your history and any pain.

For these, a side-on video you review yourself, or send to a coach, is still the better tool. Use the app for the big, visible checks and your own eyes for the details.

What types of AI form correction apps are there?

As of November 2026, the options fall into two main groups. These descriptions come from each company's own site or store listing.

  • Consumer workout apps. Strive's App Store listing says it uses the phone camera to correct form in real time and count reps.
  • Physio and rehab platforms. Kemtai describes AI-guided exercises with real-time corrective feedback, analysing 111 points on the body on any device with a camera. It is sold to clinics, health platforms and wellness apps.

The MBM app, launching soon, includes Move IQ, which uses the phone camera to count reps and flag reps that miss range, alongside coach chat. Early access is at /app.

How do you choose a good AI form correction app?

Run through these questions before you commit:

  • Does it list the exercises it supports? Vague claims like works for every exercise are a warning sign.
  • Does it tell you where to put the camera for each lift? Good apps guide your setup, because angle decides accuracy.
  • Is video processed on your phone or uploaded? Kemtai, for example, states that processing happens on the device and video is not recorded. Check the policy of any app you try.
  • Are cues specific? Go lower or knees out is useful. Bad form is not.
  • Can you reach a human? An app that pairs automated checks with coach access covers the gaps that software misses.
  • Does it work in your setting? Home workouts with clear floor space suit camera apps. Crowded gyms with racks and mirrors make tracking harder.

How do you get better feedback from any form app?

Give the app the best possible view. Google's ML Kit documentation says its pose detection works best with your whole body in frame, needs your face visible, and tracks one person at a time. In practice:

  • Prop the phone at hip height, side-on for most lifts.
  • Clear people and clutter out of the shot.
  • Wear fitted clothes so your knees and hips are visible.
  • Light yourself from the front or side, not from a window behind you.

Then act on one cue at a time. If the app flags shallow squats and caving knees in the same session, fix depth first and leave the rest for next week.

Should you trust an app over a coach?

Use both for different jobs. Apps are consistent: they apply the same rule to every rep. People are not always. Whiteside and colleagues (2016) found poor agreement between a certified screener's live grades and sensor-based measurements across six movement tests. But a coach can weigh your goals, injury history and how a lift feels, which no app can. If any exercise causes pain, see a GP or physio rather than relying on app feedback.

Common questions

Are AI form correction apps safe for beginners?

They are a helpful extra for beginners, especially for depth and range. Start with simple exercises like squats and push-ups, and get a human check on complex barbell lifts.

Do form correction apps need special equipment?

Most need only a phone or tablet camera and a way to stand it up. A cheap phone tripod makes setup faster and keeps the angle consistent.

Can an app correct form during heavy lifts?

Partly. Big movements like squat depth can still be tracked, but plates, racks and benches often block the view. Use video review for your heaviest sets.

Is my workout video private?

It depends on the app. Some say they process video on the device and do not store it. Read the privacy policy, and avoid filming other people in the gym.

Sources

  1. Schurr et al. (2017), International Journal of Sports Physical Therapy - Two-dimensional video analysis vs 3D motion capture
  2. Asaeda et al. (2024), Heliyon - Knee valgus angle calculation by posture estimation using machine learning
  3. Kemtai - Official website (accessed November 2026)
  4. Google ML Kit - Pose detection documentation
  5. Strive: AI Workout Coach - App Store listing (accessed November 2026)
  6. Whiteside et al. (2016), Journal of Strength and Conditioning Research - Grading the Functional Movement Screen
Work with Mac

Ready to apply this to your own program?

MBM builds 1:1 coaching programs around the exact principles above - weekly check-ins, weekly adjustments, and a plan that fits your life.

Keep reading

← More posts on the MBM blog