CoachLayer vs camera motion SDKs: these are different jobs.
If your product moment happens during the rep, buy a camera SDK. CoachLayer does not do that job. Live rep counting, form cues while the athlete moves, a skeleton overlay: that is on-device computer vision, it is a crowded and competent market, and a language model is the wrong tool for it.
CoachLayer works on the other side of the workout: what the session meant, whether the athlete should push or back off tomorrow, how the week went, what the program should become. Most apps that think they are choosing between the two are actually choosing both, or neither.
During the rep vs around the session.
| Dimension | Camera SDK (Sency, Kemtai class) | CoachLayer |
|---|---|---|
| The moment it serves | During the movement: counting, form scoring, live cues | Before and after: readiness, session analysis, weekly reports, program adaptation |
| Input | Camera stream, on device | Training and recovery data your app already has (sessions, sets, sleep, HRV) |
| Output | Counts, angles, scores, real-time events | Structured coaching text and decisions: briefs, analyses, next steps, adapted sessions |
| Core tech | On-device pose estimation | LLM endpoints behind engineered contracts and validators, benchmarked in production |
| Privacy surface | Camera and body imagery, which several jurisdictions treat as biometric data and regulate accordingly | Per-call stateless proxy over training data; nothing stored |
| Typical pricing | Per unique active user per month, pay as you go past 100 users; the published curve slides to roughly $0.10 per user at volume (August 2026 reading). | Credit-metered tiers, public, on the pricing table |
Where the surfaces overlap, and where we are behind.
CoachLayer ships a Form Check endpoint: send a recorded clip, get a technique breakdown in text. Three honest things about it.
It is not real-time
It analyzes a recorded clip after the set. If you need cues while the athlete moves, that is the camera SDKs’ job, and they are better at it than any clip-analysis approach will be.
It is our least proven endpoint
Our benchmark covers five of eight endpoints with production data. Form Check is not one of them yet, and we publish that instead of hiding it. Do not pick CoachLayer because of Form Check; pick it for the coaching brain and treat Form Check as a bundled extra.
The categories can stack
The strongest setups we see planned use both: an SDK counts and cues during the session, and the session facts then flow into analysis, briefs and program adaptation. Vendors in both lanes ship webhooks and JSON; the integration is natural.
On Kemtai: you are in the clinical or MSK-rehab lane where validation, compliance and payer conversations matter more than integration speed. That is a different buyer and regulator than consumer fitness, and we deliberately do not compete there.
Already have the reps counted?
Then the interesting half is what they meant. That is the layer CoachLayer sells.
Get a sandbox key