The Science Behind PacecraftAI: Our Coaching Methodology

Our Coaching Philosophy

PacecraftAI is built on the principle that elite coaching methods should be accessible to every runner — not just those who can afford a professional coach at £100–200/month. The same training science that guides Olympic marathon runners applies to the recreational runner targeting their first sub-4-hour finish. We have made it available 24/7, at a fraction of the cost.

We do not use proprietary or black-box training systems. Every coaching decision PacecraftAI makes is rooted in peer-reviewed sports science and can be explained to you in plain language.

Training Load Monitoring: ACWR and Periodisation

The foundation of our injury prevention and performance optimisation is Acute:Chronic Workload Ratio (ACWR) monitoring — the same framework used by elite athletics programmes worldwide.

ACWR compares your recent training load (the past 7 days) to your chronic load (the past 28-day average). Research by Tim Gabbett and others shows that an ACWR between 0.8 and 1.3 represents the "sweet spot" for adaptation — enough stimulus to improve without overloading the body.

• ACWR below 0.8: Under-training — adaptation rate slows
• ACWR 0.8–1.3: Optimal training zone — adaptation without excess injury risk
• ACWR 1.3–1.5: Elevated risk — monitor closely
• ACWR above 1.5: High injury risk — reduce load immediately

Polarised Training: The 80/20 Principle

The research on training intensity distribution is clear: elite endurance runners spend approximately 80% of their training volume at low intensity (conversational pace, Zone 2) and only 15–20% at moderate-to-high intensity.

This counterintuitive finding — that elite athletes run easy most of the time — has been replicated across multiple studies by Stephen Seiler and others. The mechanism: high-volume easy running builds aerobic base and mitochondrial density without generating the systemic fatigue that prevents quality hard sessions from being performed at full effort.

PacecraftAI applies the 80/20 split by default in all training plans. It will tell you if your logged runs show you running too hard on easy days (the most common mistake among recreational runners).

Periodisation: Planning the Training Year

All PacecraftAI marathon plans follow a four-phase periodisation model:

• Base phase: High volume, low intensity. Build aerobic capacity and connective tissue tolerance.
• Development phase: Introduce quality sessions (tempo, marathon-pace). Volume maintained.
• Peak phase: Highest volume and specific quality. Longest long runs. Greatest injury risk window.
• Taper phase: Volume reduced 30–50%. Intensity maintained. Body repairs and glycogen stores maximise.

Lactate Threshold and VO2 Max Development

The two primary physiological determinants of marathon performance are:

• Lactate threshold (LT2): The pace at which lactate accumulation exceeds clearance — approximately half marathon effort for trained runners. Raised by tempo runs and cruise intervals.
• VO2 max: Maximum oxygen uptake. Raised by high-intensity interval sessions (800m–1600m efforts at 3K–5K pace) and long runs that deplete glycogen.

PacecraftAI includes both LT2-targeting sessions (tempo runs) and VO2-max sessions (intervals) in all training plans above the beginner level. The frequency and volume of each are calibrated to the phase of training and the runner's event goal.

Injury Risk: What We Monitor

Beyond ACWR, PacecraftAI monitors several additional risk factors:

• Running cadence: Low cadence (<160 spm) combined with high mileage correlates with elevated knee and hip injury risk. We flag this and provide cadence coaching.
• Heart rate variability (HRV): For runners with connected devices (via Health Connect or Apple Health), below-baseline HRV signals autonomic nervous system stress and upcoming illness or overtraining. We recommend session reduction when HRV drops 15%+ below rolling baseline.
• Resting heart rate (RHR): Elevated RHR (+5–7 bpm above baseline) for multiple consecutive days is an early warning of illness or insufficient recovery. We flag this in the Health & Recovery dashboard.
• Sleep quality: Below-baseline sleep duration is associated with increased injury risk (Milewski et al., 2014 — showed <8 hours sleep tripled sports injury rate in adolescent athletes; similar effects in adult runners). We incorporate sleep data when available.

The AI System: How Coaching Decisions Are Made

PacecraftAI uses a multi-agent AI system (built on LangGraph) powered by Claude (Anthropic) as the primary language model. Each conversation goes through a routing layer that classifies the intent, passes to the appropriate specialist agent (coaching, analysis, planning, or scheduling), and synthesises the response.

Critically, the AI does not operate with static rules — it reasons about your specific situation using your actual training data, your race goals, your injury history, and the coaching science context loaded into each agent's system prompt. It can explain every recommendation it makes in the same way a human coach would.

The system never fabricates past runs or metrics you have not logged. If it does not have data to answer a question confidently, it says so and asks you to provide it.

What We Do Not Do

In the interest of transparency:

• We do not provide medical diagnosis or treatment recommendations. We flag injury risk and recommend rest or professional consultation.
• We do not replace a sports physiotherapist, doctor, or qualified human coach for complex medical or biomechanical issues.
• We do not use your data to train our models without your consent.
• Our training plan generation is based on established sports science principles, not proprietary black-box algorithms that we cannot explain to you.

References and Further Reading

• Gabbett, T.J. (2016). The training-injury prevention paradox. British Journal of Sports Medicine, 50(5), 273–280.
• Seiler, S. (2010). What is best practice for training intensity and duration distribution in endurance athletes? International Journal of Sports Physiology and Performance, 5(3), 276–291.
• Billat, V.L. (2001). Interval training for performance. Sports Medicine, 31(1), 13–31.
• Milewski, M.D. et al. (2014). Chronic lack of sleep is associated with increased sports injuries in adolescent athletes. Journal of Paediatric Orthopaedics, 34(2), 129–133.