Compare
Gloom Scroller vs Pushscroll.
How we differ. What we share. Where they win.
Gloom Scroller and Pushscroll are both apps that gate social media access on exercise, but Pushscroll uses AI pose detection via the camera to count pushups, squats, and 40+ other exercises while Gloom Scroller measures elevated heart rate via a camera fingertip pulse check.
Pushscroll counts reps; we measure effort. Same exercise-gated category, different definitions of “verified.”
Side by side
| Gloom Scroller | Pushscroll | |
|---|---|---|
| Pricing | $1.99/wk · $4.99/mo · $24.99/yr | ~$20/yr |
| Free tier | 1 blocked app, all exercises, full pulse check, streak tracking | limited; full access requires subscription |
| Core mechanic | 60-second movement interrupt + camera-based fingertip pulse check | AI pose detection (rep counting) gates app access |
| Platform | iOS only | iOS + Android |
| Hardware required | phone (rear camera + flashlight) | phone camera (pose detection) |
| Privacy posture | heart-rate data stays on-device; no cloud HR | on-device pose detection, no server processing claimed |
Pricing accurate as of May 2026.
What Pushscroll does
Pushscroll uses on-device AI pose detection to count pushups, squats, lunges, planks, and 40+ other exercises via the camera. Specific reps unlock specific minutes — five pushups equals five minutes of social media, ten squats equals ten. The exercise menu is visible and structured.
Pushscroll runs on both iOS and Android, which we don't. The pose-detection model is on-device — video doesn't leave the phone. The product targets users with an existing workout vocabulary; the structured rep-counting frame feels like an extension of an existing routine.
What we do differently
Gloom Scroller measures effort, not specific movements. There 's no proper-form requirement: 60 seconds of any movement that gets the heart rate up, then the camera reads the fingertip pulse to confirm. The bar is reachable in pyjamas, at a desk, in a hallway. No counting reps.
Different failure modes too. Pushscroll has to count reps correctly; we have to detect a pulse change. Asymmetric error tolerance is the design choice — false pass means the user still moved (the interrupt worked); false fail is solved by heart lives + skip. Our model targets non-exercisers; Pushscroll targets users who already exercise.
Last updated .