Proprietary Technology
Architecture & Algorithms
This document describes the core patentable innovations powering the TUF Axiom athletic intelligence platform. Each system represents a novel combination of AI, sports science, and real-time biometric data processing not currently available in any competing product.
Competitive Moat Summary
No Static Templates
Every workout is dynamically generated from a 2,400-config matrix. Competitors use fixed template libraries.
Phase-Aware Scoring
AXIOM scoring thresholds shift with the competition calendar. No competitor adjusts scoring gates to training phase.
CV + Scientific DB Fusion
Two-stage macro scanning blends AI vision with 330K USDA records. No sports app combines both pipelines.
Technical Specifications
The AXIOM Score Engine is a proprietary multi-dimensional readiness state machine that synthesizes six independent biometric and behavioral input streams into a single unified 0–100 readiness score. Unlike traditional single-variable fatigue trackers, AXIOM applies a weighted, phase-gated normalization algorithm that adjusts scoring thresholds dynamically based on the athlete's current training phase, sport, position, and periodization week.
Input Pipeline
Six independent streams are ingested and normalized to a 0–100 scale before fusion: • Player Load (Catapult GPS) — raw AU normalized against athlete's 21-day rolling baseline • HRV (WHOOP) — normalized against athlete's personal 30-day baseline, not population norms • Sleep Score (WHOOP) — weighted 1.4× during preseason and in-season phases • Fatigue Self-Report (1–10) — inverted and scaled; cross-validated against wearable discordance • Nutrition Adherence (%) — computed from SelectedMeal compliance vs. daily macro targets • Workout Compliance Rate (%) — rolling 7-day training completion ratio
Phase-Gated Normalization
Thresholds are not static. Each training phase applies a different normalization gate: • Off-Season: permissive thresholds, higher load tolerance, fatigue weighted lower • Preseason: HRV and sleep weighted higher; load penalties activate sooner • In-Season: tightest gates; any HRV drop >15% from baseline flags "caution" regardless of other inputs • Postseason / Rehab: recovery inputs dominate; training load inputs are suppressed This phase-aware normalization is a core differentiator — no competitor adjusts scoring boundaries based on competition calendar position.
State Classification
The fused AXIOM score maps to four discrete states: • Optimal (75–100): Full training load prescribed. Power and strength emphasis unlocked. • Train Light (50–74): Volume reduced 20–30%. Intensity maintained. Accessory work cut. • Recovery Focus (30–49): Active recovery protocol. No high-intensity work prescribed. • High Risk (<30): Training suspended. Recovery modalities + coach alert triggered. State transitions are hysteretic — an athlete must remain above a threshold for 2 consecutive days before upgrading state, preventing single-day noise from causing premature load increases.
Decision Rationale Engine
Every AXIOM output includes a machine-generated plain-language rationale string explaining the exact inputs and weights that drove the decision. This creates an auditable, explainable AI output — critical for team physicians and athletic trainers reviewing intervention decisions. The rationale is stored in AthleteDailyState.decision_rationale and surfaced to coaches in real time.
TUF Axiom Athletic Intelligence Platform
This document contains proprietary and confidential technical information. Distribution is restricted to authorized investors and legal representatives under NDA.
Document version
v2.6 · June 2026