Learning and careers decision
CheckForm Gymnastics
A competent developer can build a useful upload→pose→angle→rule scoring workflow using open-source pose projects, but reproducing the full product (real-time on-device coaching, broad skill coverage, vendor-tuned models and UI polish) is substantial and outside a small-scope replacement.
View on the App Store↗Not priced
No pricing page we fetched carried a figure, so there is nothing to compare against. The build side is still real.
$100one-off160 h to build
$200/mo6 h/mo upkeep
No published price to break even against.
No open-source build does this yet
Nothing published replaces this one, so a replacement starts from an empty file. Here is what it would have to cover.
What a replacement has to do
- Upload or record a gymnastics clip → run pose estimation → compute joint angles and simple deduction rules → return score, timestamped cues, and store attempt for progress tracking.
What it still won’t have
- Polished iOS app and App Store listing
- Real-time on-device Live Coaching with sub-100ms latency (Google ML Kit integration)
- Proprietary scoring tied to Google Gemini and vendor-tuned models
- Built-in drill library, Code of Points browser, and in-app Coach chat
- Encrypted cloud pipeline and vendor-managed data retention policies and support
What remains hard
- Product polish and ongoing maintenance
First-year cost
No published price
CheckForm Gymnastics does not publish a price we could read, so there is nothing to compare against. What building costs is below.
Money you would actually spend
Time you would spend
—
What you would spend
What we assumed
The verdict above measures whether you could build it. This one is only about money.
Runnable build prompt
Build a minimal CheckForm replacement as a web service and responsive mobile web UI using PostgreSQL, Python (FastAPI), a GPU inference host (Docker + NVIDIA runtime), and an open-source pose library (OpenPose or MMPose). Core features in scope: signed video upload and short trim; server-side batch pose inference pipeline that extracts 33 keypoints; joint-angle and timing calculations; a rule-based scoring engine for 6 common skills (handstand, cartwheel, roundoff, back handspring, handstand on beam, simple vault entry) that emits deductions and timestamped cues; overlay-rendering of skeleton on video and downloadable result clip; attempt storage and a simple progress history page; authentication (email) and account-scoped data deletion. Out of scope: native iOS app, live sub-100ms voice coaching, multi-discipline/full-code-of-points coverage, advanced ML model training. Require error handling, input validation, retries for failed inference, automated tests for API endpoints and scoring rules, and Dockerized deployment scripts for a single small GPU host.
How we checked
How the score was reached
- Partly verdict base52
- 3 cited sources+3
- 3/3 assessment runs agreed+4
- Evidence score59
The base comes from the verdict. Everything under it is a check that either happened or did not, and each one is a fact frozen in this record rather than a judgement made at render time - so the same evidence always produces the same number.
How scoring works →Cited sources · 3
Every page the run actually retrieved.
- official productGymnastics AI Coach CheckForm - App Store
- official productGymnastics AI Coach CheckForm - App Store
- official productGymnastics AI Coach CheckForm - App Store
Integrity checks
What held up, and what did not.
