Coach GPT App
AI-Driven Endurance Performance Framework • Unified Reporting v5.1
The Intervals ICU Coach V5 blends data-driven precision with evidence-based endurance science. It integrates objective training data (TSS, CTL, ATL, HRV, VO₂max) and subjective feedback (RPE, mood, sleep, recovery) to generate personalized load and wellness intelligence.
Powered by the Railway T2 Engine, this coach uses the Unified Reporting Framework (URF v5.1) to produce canonical weekly, seasonal, and wellness reports — each aligning training stress, metabolic efficiency, and recovery readiness. The system automatically interprets ACWR, Monotony, Strain, and Polarisation Index to balance hard and easy days dynamically.
Using these foundations, the coach continuously monitors ACWR, Polarisation Index, and Recovery Index to identify maladaptation early and apply proactive recovery or load modulation.
Whether you’re targeting a Gran Fondo, Ironman, or Marathon, this system ensures progressive overload, sustainable adaptation, and peak timing — maximizing aerobic performance while minimizing injury and non-functional overreaching.
Everything you ask — from “Run a weekly report” to “How efficient is my metabolism?” — connects directly to this intelligence engine, ensuring your insights are context-aware and based on canonical Intervals ICU data.
IMPORTANT NOTE: PLEASE be aware that beause of STRAVA API policy INTERVALS.ICU cannot share any STRAVA sourced data with third parties. This is otherwise fine for Garmin, Wahoo etc, and FIT file uploads to Intevals.ICU - please connect these to your intervals.icu profile.
intervalsicugptcoach.clive-a5a.workers.dev for secure OAuth authorization. Sign in, approve access, and return to ChatGPT.💡 Pro Tip: Add subjective logs in Intervals.icu (mood, soreness, stress) for richer coaching intelligence and better URF alignment.
Reports are generated automatically using the Unified Reporting Framework (URF v5.1). Each uses canonical Intervals.icu data and applies the Railway Tier-2 Engine for validation and consistency. These outputs ensure reproducible metrics, validated fatigue indicators, and synchronized wellness alignment.
Below are the available report types — click to expand. Scroll further to view anonymised sample reports illustrating full URF output.
Analyzes your last 7 days — load, intensity, and recovery alignment. Includes: ACWR, FatigueTrend, CTL / ATL / TSB, Polarisation, and metabolic drift indicators.
Period: Last 7 days (ending Dec 18, 2025)
Framework: Unified Reporting Framework v5.1
Dataset: Tier-2 canonical (event-only)
| Metric | Value | Status | Coaching Insight |
|---|---|---|---|
| ACWR | 1.02 | ✅ Productive | Balanced short/long load |
| Monotony | 1.15 | ✅ Optimal | Healthy variation |
| Strain | 69.8 | ✅ Optimal | Below overreach risk |
| Fatigue Trend | +6.1 % | ⚠ Accumulating | Monitor recovery |
| Recovery Index | 0.77 | ✅ Moderate | Avoid stacked hard days |
| Stress Tolerance | 2.0 | ⚠ Low | Increase load gradually |
| Polarisation | 1.00 | ✅ 80/20 | Clear intensity separation |
| FOx Efficiency | 0.64 | ✅ Optimal | Good fat oxidation |
| Date | Session | TSS | Duration | Distance |
|---|---|---|---|---|
| Dec 13 | Zwift – BaseCamp Tempo + Surges | 99 | 1 h 31 m | 50.2 km |
| Dec 15 | Zwift – Aerobic Endurance + Tempo | 63 | 1 h 11 m | 37.1 km |
| Dec 17 | Zwift Race – Hell of the North (A) | 80 | 1 h 07 m | 42.5 km |
| Dec 18 | Zwift – Deux zAlpe ATbase | 126 | 2 h 34 m | 54.7 km |
Total: 10 h 14 m · 210.6 km · 425 TSS
| Efficiency Factor | 1.9 | Improving power–HR efficiency |
| Fatigue Resistance | 0.95 | High durability |
| Z2 Stability | 0.04 | Steady aerobic pacing |
| Aerobic Decay | 0.02 | Minimal loss |
You are in a productive, well-balanced load state. Polarisation and metabolic efficiency are excellent, with fatigue accumulation within acceptable bounds. Maintain rhythm, extend endurance sessions, and consolidate gains with a recovery-focused day mid-week.
URF v5.1 · Canonical source: Intervals.icu Activities Dataset · Rendered via Railway Tier-2 engine
Summarizes the last 42 days of physiological and lifestyle markers: Recovery Index, HRV, Sleep Quality, Stress Tolerance, and Monotony.
Athlete: Joe Bloggs · CH Zurich, Switzerland
Generated: 18 Dec 2025 · Timezone: Europe/Zurich
| Metric | Current | Trend | Interpretation |
|---|---|---|---|
| Resting HR | 42 bpm | ↓ stable | Excellent efficiency |
| HRV | High | ↔ | Balanced autonomic state |
| VO₂ Max | ~60 | ↔ | Endurance maintained |
| Sleep Score | 80–85 | ↔ | Adequate recovery |
| Sleep Duration | 7h 15m | ↑ slight | Near-optimal |
| Mood / Motivation | 😊 High |
| Perceived Recovery | ✅ Good |
| Stress (non-training) | ⚖️ Moderate |
| Sleep Quality | 🌙 Good |
| Illness / Fatigue | 🚫 None |
You’re in a low-fatigue, high-form state — ideal for a controlled intensity block or event taper. Avoid extended detraining (>7 days).
URF v5.1 · Canonical source: Intervals.icu Wellness Dataset · Generated via Railway Render Container
Evaluates your 90-day training block: CTL growth, intensity distribution, long-term fatigue adaptation, and aerobic efficiency.
Framework: Unified Reporting Framework v5.1
Period: 90-day season block
Generated: 18 Dec 2025 · Timezone: Europe/Zurich
| Metric | Value | Interpretation |
|---|---|---|
| CTL (Chronic Load) | 74.4 | Solid endurance base |
| ATL (Acute Load) | 61.1 | Moderate recent fatigue |
| TSB (Form) | +13.3 | Fresh and well-recovered |
| ACWR | 0.92 | Productive load balance |
| Total Hours | 152.4 h | ~3-month training volume |
| Total Distance | 3,701.8 km | Consistent endurance work |
| Total TSS | 7,145 | Robust seasonal load |
| Week | Distance (km) | Hours | TSS |
|---|---|---|---|
| 38 | 195 | 8.6 | 413 |
| 39 | 266 | 12.8 | 510 |
| 40 | 280 | 11.7 | 637 |
| 41 | 191 | 14.4 | 452 |
| 42 | 289 | 16.6 | 795 |
| 43 | 310 | 10.7 | 520 |
| 44 | 324 | 11.0 | 595 |
| 45 | 407 | 13.9 | 722 |
| 46 | 242 | 10.0 | 471 |
| 47 | 411 | 14.8 | 702 |
| 48 | 240 | 7.9 | 444 |
| 49 | 168 | 5.1 | 200 |
| 50 | 237 | 9.3 | 414 |
| 51 | 144 | 5.5 | 270 |
Observation: Strong build through Weeks 40–47, followed by progressive fatigue relief and steady recovery toward Week 51.
Periodisation context:
Base → Build → Peak → Taper → Recovery.
Maintain ACWR ≤ 1.3 and allow ATL reduction of 30–50 % during taper phases.
The season profile is balanced, durable, and well-managed. Aerobic endurance and fatigue control are strong, with the block ending in a recovered, productive state — ideal for transition or sharpening.
URF v5.1 · Canonical source: Intervals.icu Activities Dataset · Rendered via Railway Tier-2 engine
Provides an overview of athlete benchmarks (FTP, VO₂max, TSB, form, and efficiency trends).
Framework: Unified Reporting Framework v5.1
Window: Rolling 90 days
Generated: 18 Dec 2025 · Timezone: Europe/Zurich
| Training Hours | 152.4 h |
| Total Distance | 3,701.8 km |
| Total Load (TSS) | 7,145 |
| Primary Sports | Ride / Run / Swim |
| Metric | Value | Interpretation |
|---|---|---|
| CTL (Fitness) | 74.4 | Solid endurance base |
| ATL (Fatigue) | 61.1 | Controlled, below risk threshold |
| TSB (Form) | +13.3 | Fresh and recovering |
Status: Form positive → ideal phase for performance or testing.
Recommendation: Maintain current load; avoid excessive tapering beyond +20 TSB.
| Measure | Status | Interpretation |
|---|---|---|
| Polarisation | ✅ Optimal (100%) | Seiler 80/20 balance confirmed |
| Load Intensity Ratio | ⚠ High (2.0) | Slightly heavy on high-intensity sessions |
| Durability Index | ✅ Stable (1.00) | Long-ride structure consistent |
| Efficiency Drift | ✅ Stable (0.00%) | Fatigue resistance sustained |
Actionable takeaway: Maintain high Z1–Z2 volume and slightly reduce mid-threshold intensity to improve aerobic economy.
You are in a productive, controlled training phase. CTL is stable, ATL well managed, and fatigue reduction signals readiness for performance testing or block transition. Focus next on Z2 duration and maintaining winter consistency.
URF v5.1 · Canonical source: Intervals.icu Activities + Wellness · Rendered via Railway Tier-2 engine
Click any question below to reveal what kind of insight it unlocks within your Intervals.icu Unified Reporting Framework (URF v5.1) analysis.
Analyzes ACWR, CTL, ATL, TSB to determine whether you’re in a recovery, productive, or overload phase.
Evaluates FatigueTrend (%) — positive means accumulating fatigue, negative means recovery trend.
Reviews the Polarisation Index to confirm zone balance (Z1/Z2 vs Z3).
Summarizes Recovery Index, HRV, Sleep, Stress, Mood, and TSB interactions to evaluate recovery alignment.
Evaluates readiness based on Form (TSB), Recovery Index, and fatigue trend.
Flags rising ATL, Stress Tolerance < 2, Monotony > 2.5, or negative HRV changes.
Breaks down FatOx Efficiency, FOxI, MES, GR, CUR to assess aerobic vs glycolytic balance.
Explains aerobic stability and cardiac drift % — <5 % = durable base; rising values = fatigue onset.
Uses CTL/ATL/TSB ratios to suggest appropriate distribution of endurance, tempo, and intensity sessions.
Generates a complete session report (power, HR, metabolic data, load impact, recovery requirement).
Each model contributes to the Coach’s adaptive intelligence system and ensures consistent endurance data interpretation:
| Model Reference | Framework Link | Metric Source | Output Type | Coaching Role |
|---|---|---|---|---|
| Seiler Polarisation | Intensity Framework | Z1–Z3% | Polarisation Ratio | Validates 80/20 intensity distribution |
| Banister Fitness–Fatigue | Load Adaptation | ATL, CTL, TSB | Training Load Model | Predicts fatigue vs adaptation trajectory |
| Coggan Power–Duration | Efficiency Framework | FTP, Power Curve | Efficiency Factor | Tracks metabolic endurance and fatigue resistance |
| Foster Overtraining | Recovery Alignment | Strain, Monotony | Overtraining Index | Detects excessive cumulative stress |
| San Millán Metabolic | Metabolic Efficiency | FatOx Index | Mito Efficiency | Evaluates fat utilization and Zone 2 economy |
| Noakes Central Governor | Readiness Forecast | HRV × RPE | CNS Fatigue Index | Detects neural fatigue and motivational readiness |
| Skiba Critical Power | Performance Integration | CP, W′ | Fatigue Decay Curve | Predicts endurance performance limits |
| Mujika Tapering | Periodisation | Load % Reduction | Taper Efficiency | Optimizes pre-event tapering blocks |
| Friel Training Stress | Consistency Framework | TSS, Compliance | Adherence Score | Validates plan execution and load control |
| Sandbakk–Holmberg Integration | Action Generation | Multi-framework synthesis | Adaptive Action Score | Produces holistic, actionable coaching feedback |
The Coach V5 functions as a multi-framework inference engine — blending classical endurance physiology with modern adaptive data modeling to intelligently evolve training plans in real time.
The Coach GPT V5 engine continuously monitors a multidimensional suite of 39–43 coaching markers across physiological, psychological, and metabolic domains. These metrics are structured into three tiers within the Unified Reporting Framework (URF v5.1).
Primary load, recovery, metabolic, and readiness metrics — continuously monitored across all report types.
| Domain | Markers | Purpose |
|---|---|---|
| Load & Performance | CTL, ATL, TSB, TSS, ACWR, Monotony, Strain, LIR, Fatigue Trend | Measure training load, balance, and adaptation trends. |
| Wellness & Recovery | Recovery Index, HRV, Resting HR, Sleep Score, Sleep Duration, Stress Tolerance, Mood, Soreness | Assess physiological recovery and psychological readiness. |
| Metabolic Efficiency | FatOx, MES, EF, Efficiency Drift %, Fatigue Resistance, Z2 Stability | Evaluate aerobic durability and energy system balance. |
| Periodisation | Block Phase, Taper Efficiency, Consistency Score, Durability Index, Adaptation Ratio | Track training phase progression and long-term load control. |
| Readiness & CNS | CNS Fatigue Index, Motivation Stability, Readiness Forecast | Measure neural recovery and performance readiness. |
| Holistic Actions | Adaptive Action Score, Action State 🟢🟠🔴, Trend Confidence % | Generate adaptive feedback and actionable coaching guidance. |
Derived from device integrations (Garmin, HRV4, Whoop) or advanced load models; used to refine URF analytics.
| Marker | Source / Dependency | Purpose |
|---|---|---|
| VO₂ Estimation (VO₂eff) | Garmin or power–HR model | Cardiorespiratory efficiency trend |
| Intensity Factor (IF) | Power data (FTP defined) | Session intensity relative to threshold |
| Session RPE (sRPE) | Manual / subjective input | Perceived load calibration (TSS × RPE) |
| Sleep Debt Index | Wellness logs (Sleep vs Target) | Quantifies chronic recovery deficit |
| HRV-SDNN / RMSSD | HRV4Training / Whoop API | Autonomic variance for deeper readiness precision |
| Glycogen Depletion Score | Power × Duration × IF model | Estimates carbohydrate utilisation |
| Hydration Score | Body weight & HR trend | Detects dehydration or plasma volume shifts |
Emerging research metrics under testing within the URF framework — included when sufficient longitudinal data exists.
| Marker | Prototype Model | Use Case |
|---|---|---|
| Running Economy Index | Stride Power / HR regression | Energy cost of running (per pace) |
| Swim Stress Coefficient | Swim TSS model (velocity × RPE) | Quantifies swim load for triathletes |
| Body Composition Drift | Weight log × Load × Recovery Index | Detects energy imbalance / under-recovery |
| Glycogen Repletion Efficiency | Post-session HR recovery × nutrition logs | Models carbohydrate uptake rate |
🧩 Total Monitored Markers: Tier-1 (32) + Tier-2 (7) + Tier-3 (3-4) → ≈ 39–43
Weighted dynamically across report types (Weekly • Seasonal • Wellness • Summary) via the URF adaptive relevance model.
The Coach GPT App is fully dependent on Intervals.icu for athlete data. All workouts, wellness logs, and training load calculations are sourced directly from your Intervals.icu account.
For best accuracy, ensure your wellness markers are synced: HRV, Resting HR, Sleep, Mood, Stress, and Soreness. Garmin users should expose VO₂maxGarmin, Performance Indicators, and Intensity Factor (IF).
🙏 Special thanks to David Tinker, creator of Intervals.icu, for enabling open endurance data access and seamless athlete integration.
Search in the GPT Store for “Intervals ICU Coach V5” or click on top link below 👇
Note: Version 3 is now deprecated — please use the latest V5 Railway Engine build for best performance and accuracy.
For integration, customization, or coaching inquiries, connect via GitHub link below or DM via Intervals.icu DM and contribute in Intervals.icu Forum.
github.com/revo2wheels
Built with ❤️ for endurance athletes — by Clive King.
Powered by Intervals.icu, Cloudflare and the Railway Engine.