NotSquat Master Training Manual & Enterprise Operating Blueprint
The long-term engineering and business blueprint for NotSquat. The current product is narrower: a privacy-preserving gym operations pilot focused on occupancy, utilization, dwell time, and congestion.
Company Identity, Executive Summary & Core Thesis
Transforming passive CCTV security feeds into the definitive real-time spatial & biomechanical data warehouse for commercial fitness, combined with gamified sprite-based member progression.
- Organization: Nerd AI (`nrd-ai`)
- Platform Entity: NotSquat Analytics LLC
- Founders & Architects: Josh Johnson (@murderszn, System & Data Architect) & Maurice Cadenhead (@mcade, Lead Vision Engineer)
- Production Portals: notsquat.org | github.com/nrd-ai/Not-Squat
- Data Warehouse: github.com/nrd-ai/NotSquat-Data-Warehouse
Commercial gym facilities invest heavily in passive IP security cameras. NotSquat is testing whether that existing infrastructure can produce useful operational metrics—without turning unvalidated estimates into customer promises.
Meanwhile, gym operators suffer from 50% annual member churn, $14.2B in annual musculoskeletal strain claims, and massive floor bottlenecks where members wait 22 minutes for squat racks while cardio decks sit 60% empty.
The Dual-Sided Value Flywheel
1. B2B Enterprise Gym Operators
- Real-time Spatial Occupancy: Automated polygon zone heatmaps, rack wait queue projections, and foot-traffic density.
- Equipment Utilization & Dwell Times: Machine-by-machine active vs idle duration tracking to inform CapEx floor planning.
- Actuarial Injury Mitigation: Real-time alerts for dangerous spinal flexion, severe knee valgus buckle, and dropped barbells.
- Future member experience: Passive member verification and workout attribution remain deferred until the operator analytics pilot, privacy review, and customer demand justify them.
2. B2B2C Member Gamification
- Zero Manual Typing: Reps, weight plate loads, cadence, and range-of-motion are automatically logged by computer vision.
- Biomechanical Feedback: Instant joint angle scoring, eccentric/concentric phase timing, and Mean Concentric Velocity (MCV).
- 16-Bit Pixel Avatar RPG: Real-life squats, deadlifts, and bench presses translate directly into avatar Strength (STR), Agility (AGI), and Endurance (END).
- Gym Community Boss Raids: Collective tonnage lifted across the entire facility inflicts real-time damage against monthly monsters on gym video walls.
4. The 20-Year Master Horizon: The Palantir of Physical Space Metadata
While commercial gyms represent our immediate beachhead, NotSquat's ultimate 20-year objective is to become the universal video-to-metadata processing engine across all sleeping giants of passive surveillance (gyms, train stations, airports, commercial banks, government and manufacturing facilities).
[ Phase 1: Fitness & Biomechanics (2026-2029) ]
• 10,000+ commercial gym CCTV feeds ingested via zero-cost data barter.
• Unfair foundation AI moat for dense human kinematics under extreme occlusion & mirrors.
[ Phase 2: Transit Hubs, Banking & Commercial Retail (2030-2035) ]
• Train stations (Amtrak, MTA, BART), airports, and bank branches.
• Zero-PII spatial flow rates, queue saturation curves, and crowd safety alerts.
[ Phase 3: Universal Physical Space Palantir (2035-2046) ]
• The private-sector Palantir where enterprise operators query physical space like a database:
SELECT avg(dwell_time), queue_wait FROM transit_hubs WHERE station_id = 'GRAND_CENTRAL';
We do not conduct surveillance for the sake of intrusive identity monitoring. We conduct edge video processing for the sake of structured, privacy-safe metadata extraction that unlocks operational intelligence, safety risk reduction, and gamified human achievement.
The "Data-for-Intelligence" Barter Flywheel & Commercial Pitch
Our consent-first gym operations pilot, validation milestones, future monetization hypotheses, privacy boundaries, and long-term expansion notes.
1. The Zero-Cost Gym Pilot Barter Strategy
Instead of cold-pitching conservative gym operators with high software subscription fees, NotSquat executes a frictionless zero-cost pilot exchange:
[ Partner Gym Operator ] ──► (Provides Archival / Sample CCTV Feeds) ──► [ NotSquat Engine ]
│
┌───────────────────────────────────────────────────────────────────────┘
▼
[ Gym Receives 100% FREE Enterprise Data Warehouse & Analytics ]
• Comprehensive Machine & Squat Rack Utilization Heatmaps
• Peak Demand Bottleneck & Queue Wait Duration Projections
• Biomechanical Safety & Actuarial Injury Violation Flags
• Predictive Sanitation Alerts (triggered every 15 uses)
│
▼
[ NotSquat Retains De-Identified Biomechanical Training Vectors ]
• Ground-truth fitness movement dataset across crowded, reflective commercial gym environments
• Unfair computer vision moat for plate recognition, bar velocity, and obscure movement kinematics
2. Enterprise SaaS Pricing Ladder & Monetization Streams
For single-facility independent gym owners seeking floor telemetry.
- Facility spatial heatmaps
- Machine dwell & idle metrics
- Automated sanitation triggers
- Weekly operational PDF digests
For high-traffic strength facilities requiring safety risk reduction.
- Everything in Core Operations
- Real-time dropped barbell alerts
- Critical knee valgus & spinal flags
- Un-racked weight hazard detection
- Actuarial insurance audit logs
The complete enterprise suite with consumer gamification app.
- Everything in Safety & Liability
- Full 16-Bit Pixel Member App
- Zero-friction auto rep logging
- Community Boss Raid TV Wall Engine
- ABC / Mindbody CRM 2-way sync
Additional Revenue Engines:
- High-Ticket T&M Consulting ($300 – $450/hr): Custom branded member mobile apps, proprietary turnstile hardware integrations, and on-premise NVIDIA Jetson server rack deployment.
- Commercial Biomechanical Foundation Model Licensing: Licensing de-identified movement kinematics to fitness hardware manufacturers (Life Fitness, Rogue, Hammer Strength) and insurance actuaries.
3. Interactive Gym ROI & Profit Expansion Calculator
Explore the monthly and annualized bottom-line impact of deploying NotSquat across standard commercial fitness facilities:
4. 3-Tier Privacy Objection-Handling Script
| Privacy Tier | Core Security Protocol | Executive Objection Response Script |
|---|---|---|
| Tier 1: Ephemeral Local Processing | RAM-only ingestion; raw video frames discarded within 100ms; zero disk writes. | "We never save your CCTV video. Our edge appliance processes pixels in volatile memory, extracts 33 mathematical joint coordinates, and immediately deletes the video stream." |
| Tier 2: Automated Face & PII Blurring | Hardware-level Gaussian facial blurring applied before any analytical telemetry aggregation. | "Every face and license plate is irreversibly blurred in hardware before spatial heatmaps or machine dwell stats are tabulated. Only opted-in members are recognized." |
| Tier 3: Zero Sensitive Zones Policy | Hardware and software isolation excluding locker rooms, bathrooms, saunas, and private suites. | "NotSquat is strictly hardcoded for public workout floor areas. We enforce a zero-tolerance exclusion policy on all locker room and private zones." |
5. One-Page Gym Operator Pilot Proposal Template (MOU)
MEMORANDUM OF UNDERSTANDING: NOTSQUAT AI FACILITY PILOT PROGRAM
BETWEEN: NotSquat Analytics LLC ("NotSquat")
AND: [Partner Gym Facility Name] ("Partner Gym")
1. PURPOSE & SCOPE:
NotSquat shall provide Partner Gym with full access to the NotSquat Spatial & Biomechanical
Data Warehouse at ZERO ($0.00) software licensing cost. In exchange, Partner Gym authorizes
NotSquat to process sample security camera footage strictly for metadata extraction, spatial
optimization, and model training.
2. DELIVERABLES PROVIDED TO PARTNER GYM (100% FREE):
- Comprehensive Machine & Squat Rack Utilization Heatmaps.
- Peak Demand Bottleneck & Queue Wait Duration Analysis.
- Safety & Biomechanical Form Diagnostics (Valgus collapse / injury flags).
- Custom Executive Facility Report detailing spatial floor optimization opportunities.
3. PRIVACY & DATA ETHICS COMMITMENT:
- All video streams are processed under strict RAM-only anonymization protocols.
- No biometric identity vectors or raw video files are sold to third parties.
- Facial regions are irreversibly blurred prior to analytical archiving.
4. COMMERCIAL EXTENSION OPTIONS (POST-PILOT):
Upon review of the metadata, Partner Gym may optionally engage NotSquat for:
a) Turnkey Member Gamified Mobile App (Custom Development @ $350/hr).
b) 24/7 Real-Time Edge Video Telemetry & Alerting SaaS ($399 - $999/mo).
AGREED & ACCEPTED:
For NotSquat Analytics LLC: ______________________ Date: ___________
For Partner Gym Facility: ______________________ Date: ___________
15-Slide Executive Pitch Deck Walkthrough
The institutional-grade board presentation tailored for executive leadership at Gold's Gym, Planet Fitness, Equinox, Life Time, and Crunch Fitness.
End-to-End Computer Vision & Biomechanical Math Architecture
33 3D skeletal landmark extraction, vector geometry, knee valgus ratio calculations, optical flow velocity, and sub-15ms edge inference.
1. 3D Keypoint Landmark Geometry
NotSquat captures spatial-temporal joint coordinates $P_i = (x_i, y_i, z_i)$ with visibility confidence scores $c_i \in [0.0, 1.0]$. Given three sequential anatomical keypoints $P_1$ (proximal), $P_2$ (joint vertex), and $P_3$ (distal):
$$\vec{u} = P_1 - P_2 = \begin{pmatrix} x_1 - x_2 \\ y_1 - y_2 \\ z_1 - z_2 \end{pmatrix}, \quad \vec{v} = P_3 - P_2 = \begin{pmatrix} x_3 - x_2 \\ y_3 - y_2 \\ z_3 - z_2 \end{pmatrix}$$
$$\theta = \arccos\left(\frac{\vec{u} \cdot \vec{v}}{\|\vec{u}\|_2 \, \|\vec{v}\|_2}\right) \times \left(\frac{180^\circ}{\pi}\right)$$
2. Biomechanical Metric Mathematical Definitions
| Biomechanical Metric | Mathematical Formula | Operational Significance |
|---|---|---|
| Knee Flexion Angle ($\theta_{\text{knee}}$) | $\angle(\text{Hip}_{23/24}, \text{Knee}_{25/26}, \text{Ankle}_{27/28})$ | Validates true squat depth ($70^\circ - 95^\circ$ parallel/below) vs quarter-rep high squats ($>110^\circ$). |
| Knee Valgus Ratio ($V_R$) | $$V_R = \frac{\|\text{Knee}_L - \text{Knee}_R\|_2}{\|\text{Ankle}_L - \text{Ankle}_R\|_2 + \epsilon}$$ | Detects inward medial knee buckling. Ratio $< 0.75$ or inward collapse $> 12^\circ$ flags severe ACL tear risk. |
| Spinal Neutrality ($\Delta \theta_{\text{spine}}$) | $$\Delta \theta = |\angle(\text{Shoulder}_{\text{mid}}, \text{Hip}_{\text{mid}}) - \theta_{\text{baseline}}|$$ | Monitors lumbar flexion ("butt wink"). Angular variance $> 12^\circ$ under load triggers critical injury alerts. |
| Mean Concentric Velocity (MCV) | $$\text{MCV} = \frac{1}{T_c} \int_{t_{\text{inflection}}}^{t_{\text{lockout}}} \frac{dy_{\text{bar}}}{dt} \, dt$$ | Measures explosive speed in meters/second. Quantifies velocity loss percentage to estimate RPE and fatigue. |
| Re-ID Appearance Vector | $$\vec{f}_{\text{ReID}} = \text{FastReID}(\text{Crop}_{\text{person}}) \in \mathbb{R}^{512}$$ | 128-D to 512-D normalized HSV appearance embedding for non-overlapping multi-camera member handover. |
3. Edge Latency Budget Breakdown
| Pipeline Stage | Algorithm / Engine | Hardware Target | Latency |
|---|---|---|---|
| Hardware RTSP Decode | NVDEC / GStreamer H.264/H.265 | Jetson AGX Orin / RTX 4090 | 1.8 ms |
| Person & Barbell Detection | YOLOv11x (TensorRT FP16) | Jetson AGX Orin / RTX 4090 | 3.2 ms |
| 33-Keypoint Pose Estimation | RTMPose-l / MediaPipe Pose | Jetson AGX Orin / RTX 4090 | 4.1 ms |
| Kinematic & Tripwire Math | Vectorized C++ / NumPy SIMD | Host CPU Core | 0.6 ms |
| Multi-Camera Re-ID Handover | FastReID + ByteTrack | TensorRT / Qdrant HNSW | 2.5 ms |
| WebSocket Telemetry Broadcast | FastAPI async / ujson | Edge Node Host | 1.1 ms |
| TOTAL END-TO-END LATENCY (Camera Lens to Member App) | 13.3 ms (< 1 frame at 60 FPS) | ||
Commercial Exercise Telemetry & Biomechanical Catalog
Exhaustive movement standards, anatomical targets, and 3-tier joint angle threshold definitions across Barbells, Dumbbells, Cables, Machines, and Calisthenics.
1. Barbell Back Squat (High Bar & Low Bar)
Free Weights • Barbell • Compound PrimaryAnatomical Targets: Agonists: Quadriceps femoris (rectus femoris, vastus lateralis/medialis/intermedius), Gluteus maximus. Synergists: Adductor magnus, Soleus, Hamstrings. Stabilizers: Erector spinae, Rectus abdominis, Obliques.
2. Barbell Front Squat
Free Weights • Barbell • Anterior Quad BiasAnatomical Targets: Agonists: Quadriceps femoris (vastus lateralis/medialis). Synergists: Gluteus maximus, Adductor magnus, Anterior Deltoids. Stabilizers: Thoracic extensors, Transverse abdominis, Rhomboids.
3. Conventional Barbell Deadlift
Free Weights • Barbell • Posterior ChainAnatomical Targets: Agonists: Gluteus maximus, Hamstrings (biceps femoris, semitendinosus), Quadriceps off floor. Synergists: Adductor magnus, Latissimus dorsi, Trapezius, Soleus. Stabilizers: Erector spinae, Forearm flexors.
4. Barbell Flat Bench Press
Free Weights • Barbell • Upper Body Horizontal PushAnatomical Targets: Agonists: Pectoralis major (sternal & clavicular heads). Synergists: Anterior deltoid, Triceps brachii. Stabilizers: Rotator cuff, Latissimus dorsi, Serratus anterior.
5. Barbell Overhead Press (OHP / Strict Press)
Free Weights • Barbell • Vertical PushAnatomical Targets: Agonists: Anterior deltoid, Lateral deltoid. Synergists: Triceps brachii, Clavicular pec, Upper traps, Serratus anterior. Stabilizers: Core musculature, Glutes.
6. Dumbbell Flat & Incline Bench Press
Free Weights • Dumbbell • Unilateral PushAnatomical Targets: Agonists: Pectoralis major. Synergists: Anterior deltoid, Triceps brachii. Stabilizers: Rotator cuff, Serratus anterior.
7. Dumbbell Lateral Raise & Standing Bicep Curl
Free Weights • Dumbbell • Isolation TelemetryAnatomical Targets: Lateral Raise: Lateral deltoid, Supraspinatus, Trapezius. Bicep Curl: Biceps brachii, Brachialis, Brachioradialis.
8. Cable Lat Pulldown & Seated Cable Row
Cable & Pulley • Upper Body PullingAnatomical Targets: Agonists: Latissimus dorsi, Teres major, Rhomboids, Middle trapezius. Synergists: Biceps brachii, Brachialis, Posterior deltoid.
9. 45-Degree Leg Press & Hack Squat Machine
Machines • Plate-Loaded Sled TelemetryAnatomical Targets: Agonists: Quadriceps femoris, Gluteus maximus. Synergists: Adductor magnus, Soleus.
10. Pull-Up, Chin-Up & Parallel Bar Dips
Calisthenics • Bodyweight KinematicsAnatomical Targets: Pull-Up: Latissimus dorsi, Biceps, Rhomboids. Dips: Triceps brachii, Lower pectoralis major, Anterior deltoid.
11. Strict Push-Up & Plank Core Stack
Calisthenics • Horizontal Bodyweight PushAnatomical Targets: Agonists: Pectoralis major, Anterior deltoid, Triceps brachii. Stabilizers: Rectus abdominis, Obliques, Quadriceps, Erector spinae.
16-Bit Gamification Engine, Sprite Mechanics & RPG Economics
Transforming physical exertion into synchronous 16-bit pixel avatar progression, powerlifting unlocks, golden PR auras, and cooperative gym boss raids.
1. Avatar Attributes & Mathematical Scaling
| RPG Attribute | Real-World Physical Stimulus | Mathematical Scaling & Level Formula |
|---|---|---|
| STR (Strength) | Heavy compound lifts (Squat, Deadlift, Bench, OHP) | $$\Delta \text{STR} = \sum \frac{\text{Weight}_{\text{lbs}} \times \text{Reps} \times \text{ROM}_{\%}}{1,000}$$ |
| AGI (Agility) | Explosive bar speed, Mean Concentric Velocity, HIIT | $$\Delta \text{AGI} = \text{MCV (m/s)} \times \text{Cadence Rate Bonus}$$ |
| END (Endurance) | High-rep volume (>12 reps), minimal rest intervals, cardio | $$\Delta \text{END} = \text{Active Minutes} + \left(\frac{\text{High-Rep Volume}}{500}\right)$$ |
| DISC (Discipline) | Streak continuity, strict form adherence (no yellow/red flags) | $$+5\% \text{ per consecutive day}, +15\% \text{ per 100-score workout}$$ |
2. Experience (XP) Dynamic Equation & Level Curve
$$\text{XP}_{\text{set}} = \left(\text{Reps} \times \frac{\text{Weight}_{\text{lbs}}}{10}\right) \times \left(\frac{\text{Form Score}}{100}\right) \times \text{Streak Multiplier}$$
$$\text{Required Cumulative XP}(L) = 1,000 \times L^{1.65}$$
3. Gym Community Boss Raids
Every month, each partner facility spawns an automated 16-Bit Cyber Behemoth on the main facility video wall (e.g. "The 1,000,000 LB Iron Titan"). Every rep completed inside the club inflicts real-time damage: 1 lb lifted = 1 HP damage. Defeating the boss unlocks limited-edition sprite cosmetic auras and smoothie bar rewards.
MongoDB Star Schema Data Warehouse & Processing Engine
Raw edge ingestion models, Star Schema dimensions, telemetry fact tables, and Velocity-Based Training (VBT) 1RM linear regression profiling.
Open source repository: `nrd-ai/NotSquat-Data-Warehouse`.
{
"frame_id": "frm_a819b2c3-4d5e",
"timestamp": "2026-08-17T21:15:00.120Z",
"camera_id": "CAM_RACK_01",
"facility_id": "FAC_AUSTIN_01",
"track_id": 4,
"bounding_box_2d": { "x_min": 0.35, "y_min": 0.15, "x_max": 0.65, "y_max": 0.85 },
"optical_flow_vector": { "dx": 0.02, "dy": -0.45, "magnitude": 0.45 },
"landmarks_3d": [
{ "landmark_id": 0, "name": "NOSE", "x": 0.50, "y": 0.22, "z": -0.15, "visibility": 0.98 },
{ "landmark_id": 23, "name": "LEFT_HIP", "x": 0.46, "y": 0.52, "z": 0.02, "visibility": 0.95 },
{ "landmark_id": 25, "name": "LEFT_KNEE", "x": 0.44, "y": 0.70, "z": 0.08, "visibility": 0.96 },
{ "landmark_id": 27, "name": "LEFT_ANKLE", "x": 0.45, "y": 0.88, "z": 0.05, "visibility": 0.97 }
]
}
{
"rep_id": "rep_948a1b2c",
"session_id": "SES_20260817_01",
"member_key": "MBR_001",
"exercise_key": "EX_SQUAT_BACK",
"station_key": "STATION_RACK_01",
"camera_key": "CAM_RACK_01",
"date_key": 20260817,
"set_index": 3,
"rep_index": 5,
"load_lbs": 315.0,
"duration_seconds": 2.85,
"eccentric_seconds": 1.85,
"concentric_seconds": 1.00,
"mean_concentric_velocity": 0.58,
"peak_velocity": 0.74,
"knee_flexion_rom": 118.4,
"hip_depth_status": "BELOW_PARALLEL",
"spinal_neutrality_score": 94.5,
"valgus_collapse_ratio": 0.04,
"rpe_estimated": 7.5,
"form_score": 96.0,
"xp_earned": 45
}
{
"snapshot_id": "OCC_FAC_AUSTIN_01_ZONE_POWER_RACKS_20260817_181500",
"facility_key": "FAC_AUSTIN_01",
"zone_key": "ZONE_POWER_RACKS",
"date_key": 20260817,
"timestamp": "2026-08-17T18:15:00.000Z",
"headcount": 18,
"occupied_stations": 6,
"total_stations": 6,
"queue_length": 2,
"utilization_rate": 1.00,
"avg_dwell_minutes": 22.4
}
{
"member_key": "MBR_001",
"name": "Marcus Aurelius Vance",
"membership_tier": "VIP_ALL_ACCESS",
"home_facility_key": "FAC_AUSTIN_01",
"height_cm": 182.5,
"bodyweight_lbs": 198.4,
"rpg_guild": "IRON_LEGION",
"rpg_class": "HEAVY_PALADIN",
"level": 14,
"total_xp": 18450,
"biometric_opt_in": true,
"is_active": true
}
Velocity-Based Training (VBT) 1RM Linear Regression Profiling
The warehouse analytics engine automatically computes a lifter's actual 1RM by fitting a linear regression line across load vs. Mean Concentric Velocity (MCV) data points, solving for the Minimum Velocity Threshold ($MVT = 0.30\text{ m/s}$ for back squat):
$$\text{MCV} = m \cdot \text{Load} + b \implies \text{Estimated 1RM} = \frac{MVT - b}{m}$$
Legal, BIPA, GDPR Article 9 & CCPA Privacy Compliance Blueprint
Regulatory safe harbor architectures, zero-storage edge ring buffers, cryptographic mTLS security, and automated right-to-be-forgotten protocols.
Biometric facial templates are collected strictly opt-in via written digital release embedded in member apps. Templates are permanently destroyed upon account termination or after 36 months of inactivity. Zero biometric data is ever sold or shared.
Special category biometric data is processed on-premise with explicit consent. Implements Article 17 Right to Erasure (1-click automated purge of all appearance vectors) and Article 25 Privacy by Design (pseudonymized Re-ID tokens).
Strict "Do Not Sell My Information" compliance. Gym members can request a complete, instant machine-readable export of all historical joint angles, rep logs, and workout telemetry at any time.
Cryptographic Transport Security
- TLS 1.3 with AES-256-GCM: All edge-to-cloud streams are strictly encrypted in transit.
- Hardware mTLS (Mutual TLS): On-premise Jetson nodes authenticate using X.509 client certificates issued by NotSquat Internal CA.
- HMAC-SHA256 Signatures: Every WebSocket telemetry frame carries facility-scoped signatures to prevent replay attacks or packet tampering.
Open-Source SOTA AI Model Stack & Benchmarks
Curated foundation models, pose estimation benchmarks, spatio-temporal action networks, and vector search infrastructure.
| Framework / Engine | Open Source Repository | Model Architecture | Inference Speed / Edge Role |
|---|---|---|---|
| MMPose (RTMPose) | `open-mmlab/mmpose` | RTMPose-l (SimCC keypoint representation) | 100+ FPS on edge GPU; 33 3D landmarks |
| Ultralytics YOLO-Pose | `ultralytics/ultralytics` | YOLOv11-Pose (17 COCO Keypoints) | 3.2 ms on TensorRT FP16; instant person detection |
| Google MediaPipe | `google/mediapipe` | BlazePose (33 3D full-body landmarks) | Sub-pixel depth; CPU/Edge fallback |
| MMAction2 | `open-mmlab/mmaction2` | ST-GCN / PoseC3D Temporal Graph Convs | Temporal exercise classification & rep phase gating |
| FastReID | `JDAI-CV/FastReID` | BoT-SORT / FastReID 512-D embedding | Non-overlapping multi-camera member handover |
| InsightFace (ArcFace) | `deepinsight/insightface` | ArcFace Cosine Margin Classifier | Hands-free turnstile check-in (>99.8% accuracy) |
| ClickHouse & MongoDB | `ClickHouse/ClickHouse` | Columnar & Star Schema Document Store | Sub-millisecond aggregation of millions of rep rows |
| Qdrant Vector DB | `qdrant/qdrant` | HNSW Vector Indexing | Sub-5ms Re-ID and facial vector neighbor search |
Complete REST & WebSocket API Specification
OpenAPI 3.1.0 specification, live WebSocket telemetry channels, downstream event broadcast schemas, and SDK integration examples.
1. WebSocket Live Telemetry Channels
WebSocket Host: wss://stream.notsquat.ai/ws/v1 (Protocol: notsquat-telemetry-v2)
{
"event": "REP_COMPLETED",
"timestamp": "2026-08-17T19:58:30.120Z",
"data": {
"rep_id": "rep_9482910",
"member_id": "usr_994821",
"exercise_type": "barbell_squat",
"rep_index": 4,
"metrics": {
"duration_sec": 2.45,
"concentric_sec": 1.05,
"eccentric_sec": 1.40,
"min_knee_flexion_deg": 76.4,
"valgus_ratio": 0.94,
"mean_bar_velocity_mps": 0.58,
"form_score": 96.2
},
"gamification": {
"xp_awarded": 50,
"str_gain": 0.15,
"boss_damage": 315.0,
"sprite_animation": "squat_power_sparkle"
},
"tripwires": []
}
}
2. Python Client SDK Integration
import asyncio
import websockets
import json
async def stream_notsquat_telemetry(gym_id: str, member_id: str, api_key: str):
uri = "wss://stream.notsquat.ai/ws/v1"
headers = {"X-NotSquat-Gym-Key": api_key}
async with websockets.connect(uri, extra_headers=headers) as ws:
# Subscribe to member rep stream
sub_payload = {
"action": "subscribe",
"channel": "member.rep_stream",
"params": {"gym_id": gym_id, "member_id": member_id}
}
await ws.send(json.dumps(sub_payload))
print(f"[*] Subscribed to live telemetry for member: {member_id}")
while True:
raw_msg = await ws.recv()
event = json.loads(raw_msg)
if event.get("event") == "REP_COMPLETED":
d = event["data"]
print(f"[REP {d['rep_index']}] {d['exercise_type']} | Depth: {d['metrics']['min_knee_flexion_deg']}° | Score: {d['metrics']['form_score']}% | +{d['gamification']['xp_awarded']} XP")
if __name__ == "__main__":
asyncio.run(stream_notsquat_telemetry("gym_golds_venice_01", "usr_994821", "nsq_live_demo_key"))
Pre-Client Roadmap & Operating Milestones
Tracking core execution milestones across Video Acquisition, Brand Identity, Corporate Formation, Marketing Presence, and the Autonomous AI Agent Fleet (Issues #7 through #15).
Acquire initial archival CCTV / MP4 surveillance video datasets from pilot gym partners (including Rupert's gym and Austin regional facilities). Ingest raw feeds to build ground-truth edge calibration matrices and validate bounding box tracking across dynamic gym lighting and chalk clouds.
Comprehensive market analysis benchmarking NotSquat against legacy computer vision tools, wearable sensors (WHOOP, Apple Watch), and optical tracking hardware. Formalize our zero-hardware, multi-camera Re-ID moat.
Package TensorRT-accelerated edge inference containers for dual NVIDIA Jetson AGX Orin nodes. Establish automated model evaluation harness benchmarking RTMPose vs YOLOv11-Pose across occlusion scenarios.
Finalize production design system, SVG vector badges, retro 16-bit favicon suite, and high-resolution dark-mode corporate assets for the NotSquat ecosystem.
Configure enterprise Google Workspace communication infrastructure on notsquat.org / notsquat.ai (e.g. enterprise@notsquat.ai, maurice@notsquat.ai, legal@notsquat.ai) with strict DMARC, DKIM, and SPF authentication records.
Secure unified handle @notsquatorg across X (Twitter), Instagram, YouTube, TikTok, LinkedIn, and GitHub. Deploy automated video clip syndication showcasing real-time sprite synchronizations and form analysis.
Incorporate NotSquat Analytics LLC, execute formal founder Operating Agreement between Josh Johnson and Maurice Cadenhead, and draft enterprise Master Services Agreements (MSA) with BIPA/GDPR biometric compliance indemnification riders.
Convert Markdown board decks into interactive web pitch decks and executive PDF one-pagers tailored for franchise board presentations at Gold's Gym, Planet Fitness, and Equinox.
Deploy specialized autonomous AI agents to drive continuous operations: Lead Gen Agent (scraping commercial gym operator registries), Biomechanics Research Agent (scanning sports science papers for joint angle bounds), Content Agent (generating 16-bit social workout reels), and Enterprise Outreach Agent.