Product preview — features and pricing are planned unless marked available.
Lumos

Learn to feel the muscle you're training

The mind-muscle connection is real and measurable: internal focus roughly doubled biceps growth versus external focus at moderate loads in Schoenfeld's 2018 trial (12.4% vs 6.9%). SwoleCraft turns that finding into tooling.

Citation: Schoenfeld BJ et al. (2018), Eur J Sport Sci 18(5) — every claim in this app maps to a citations registry.

Where did you feel it?

The Personal Activation Map

After key sets, one question and one tap. Over weeks, SwoleCraft builds a diary of where you report feeling each exercise. It is self-report, not an activation measurement. Try it — this demo expects a barbell row:

You just rowed. Tap where you felt it:

Front deltsSide deltsPecs (chest)BicepsForearmsAbsObliquesDeep coreAdductorsQuadsfront
Upper trapsMid/lower trapsRhomboidsRear deltsTricepsLatsSpinal erectorsGlute medGlutesHamstringsCalvesback
Faint glow = what this movement should hit. Your taps teach the map.
1

Cue change first

Report more biceps than lats? SwoleCraft can offer an internal lat cue while leaving the rest of the set unchanged.

2

Tempo option second

If you report the same mismatch again, the app can offer a slower eccentric and a modest load reduction for you to accept or reject.

3

Exercise option third

If you want another approach, the app can offer a pattern-sibling such as a chest-supported row.

The smart cue engine

Internal cues for growth. External cues for load.

The cited curl trial supports an internal-focus option at moderate loads. SwoleCraft uses short outcome cues for heavy compounds as a product choice, and labels which mode it selected. Try the scenarios:

external cue

Push the floor away.

SwoleCraft uses an outcome-focused cue here so the instruction stays simple under a heavy bar.

P1 · the Lumos centerpiece

The 3D écorché model

Every muscle a separate named mesh, mapped 1:1 to the same registry as the 2D map. Preview mode glows targets pre-set; Anatomy mode ranks each muscle's best exercises by your own activation data. The 2D map you've been tapping is the same data — the 3D model is its premium body.

Under evaluation (docs/10 §5): licensed stock écorché vs open anatomy datasets vs a commissioned clean-room model. Hard requirement either way: per-muscle named meshes (pectoralis_major_L, …), ≤15 MB glTF+Draco, 60fps orbit on a Pixel 6 — with automatic fallback to this 2D map on low-tier devices.