KI Data Capture & Interpretation iPad App — Developer Collaboration Brief 1) Overview I’m building an iPad app for Kinetic Intelligence® (KI) that captures movement performance (video + sensor data) and converts it into simple, actionable KI scores and coaching decisions in real time. This is not a generic “movement app.” It’s a pitch-side performance and rehab decision tool designed for: • Football clubs (primary) • Education/training bodies (secondary) • Performance/rehab practitioners I (Kevin Pratt) will collaborate closely as the KI methodology owner: defining metrics, interpreting outputs, designing test protocols, and validating the coaching recommendations. ⸻ 2) Project Goals (What “success” looks like) V1 Launch Goal: a polished iPad app that can be used pitch-side to: • Record video of an athlete performing specific tests • Overlay pose tracking live/on replay • Auto-segment reps (or provide easy manual trimming) • Compute and display KI metrics with confidence • Generate a “KI Snapshot” screen that outputs: • Scores (MI / LI / DI) • 3 core KI metrics with trends • One stoplight risk flag • One action plan (cue + constraint + drill + progression rule) • Export session summaries (PDF + annotated video clip) Primary objective: immediate coaching decisions, not data overload. ⸻ 3) Target Platform & Tech Preferences Platform: iPadOS (iPad-first experience) Preferred stack (open to discussion): • Swift / SwiftUI • AVFoundation (video capture) • Apple Vision pose estimation (2D; 3D where supported) • Core Motion (IMU signals for smoothness/intensity proxies) • Optional later: ARKit/RealityKit for LiDAR devices Constraints: • Offline-first (works on pitch, sync later optional) • Fast: near real-time scoring/feedback • Privacy: data stored locally by default; cloud sync can be phased in ⸻ 4) Core User Workflows (Must-have) A) Create session • Select athlete (or quick-add) • Choose session template (Decel / COD / Sprint / Landing-Reaccel / RTP Bridge) • Select intent (e.g., Soft / Fast / Stable / Powerful / Reactive) B) Capture rep(s) • Record video • Show skeleton overlay • Auto detect rep start/stop (or coach-friendly manual trim) • Tag key context quickly (surface type, fatigue, pain, RPE) C) Review + Interpret • Replay with overlay + key event markers • Display KI Snapshot • Show “Why” (top 3 drivers) and confidence score • Provide action plan (cue + constraint + drill + progression rule) D) Track • Athlete timeline with trends over time (LAT) • Compare to baseline / previous sessions • Flag regressions and plateaus E) Export • PDF summary • Shareable annotated clip (overlay + headline scores) ⸻ 5) V1 Metrics (Launch Set) App must support metric computation + explanation with confidence scores: 1. Movement Optionality Index™ (MOI) “How many usable movement solutions can the athlete access under constraint?” 2. Intent–Output Coherence™ (IOC) “Did movement output match stated intent?” 3. Adaptive Capacity Ratio™ (ACR) “How well is quality maintained as task difficulty increases?” 4. Stability–Variability Balance™ (SVB) “Is the athlete stable enough but not rigid; variable enough but not chaotic?” 5. Longitudinal Adaptation Trajectory™ (LAT) “Are scores moving in the right direction over time?” Important: I will provide definitions, scoring rules, thresholds, and test protocols. The developer builds the pipeline, data structures, and UI to compute/display these reliably. ⸻ 6) “KI Snapshot” Screen (Signature Feature) One screen that sells the system: • MI / LI / DI scores (large, simple) • MOI, IOC, ACR (with trend arrows) • One stoplight risk indicator (green/amber/red) • Action Plan auto-generated: • 1 cue • 1 constraint • 1 drill • 1 progression rule • “Show me why” expands to top drivers and annotated replay ⸻ 7) Functional Requirements Capture • Stable video recording with good UX • Frame-accurate timestamping • Save raw video + processed features Pose pipeline • Joint tracking overlay (replay + live if possible) • Confidence per joint / per rep • Smoothing + dropout handling Rep segmentation • Auto detection (v1 can be heuristic + manual override) • Coach-friendly trimming and labeling Metric engine • Modular computation per test type • Outputs: score 0–100 + confidence + drivers • Ability to update metric definitions without rewriting the whole app Data model • Athlete profile • Baselines • Sessions → tests → reps → features → metric results • Versioning (so metric updates don’t break historical data) Export • Clean PDF report generation • Export clip with overlay + headline metrics (at least basic) ⸻ 8) Non-Functional Requirements • Fast UI, minimal taps • Robust offline behavior • Data integrity + backups • Clear error states (“tracking lost”, “confidence low”) • Logging/telemetry (local debug logs at minimum) ⸻ 9) Collaboration Expectations I will provide: • KI methodology, definitions, and scoring logic • Test protocols and success criteria • Interpretation language and action plan rules • Pilot feedback from practitioners Developer/studio provides: • iPad app engineering (capture, pose, pipelines, UI) • Technical decisions + architecture • Build iterations and QA • TestFlight builds and release support Working rhythm: • Weekly sprint planning + review • Rapid iterations with me testing in the field • Shared backlog (Notion/Jira/Trello) ⸻ 10) Deliverables Phase 1 (MVP Prototype) • Capture + pose overlay + rep trim • Manual tagging • MOI + IOC computed for 1–2 session templates • Basic KI Snapshot Phase 2 (Launch v1.0) • Full launch metric set (5) • 5 football-first session templates • Export PDF + annotated clip • Athlete profiles + trends (LAT) • Polished UI/UX Phase 3 (Post-launch enhancements) • ARKit/LiDAR depth (where supported) • Core ML classification for movement types/quality tiers • Cloud sync + team dashboards ⸻ 11) What I Need From You (Developer Response) Please include: • Recommended architecture + stack • Timeline estimate by phase • Team composition (iOS, ML/CV, designer) • Relevant prior work (video, computer vision, sports/health) • Risks/constraints you foresee • Ballpark cost range per phase (if applicable) ⸻ 12) IP, Data, and Commercial Notes • KI metrics and naming are proprietary to Kinetic Intelligence® • Developer code can be negotiated (work-for-hire preferred) • Athlete data is sensitive; privacy-by-design is essential • Any third-party SDKs must be disclosed and approved