Will AI Replace Endurance Coaches? Stress Biology, Complexity Science, and the Future of Coaching

By Bevan McKinnon

About the author: Bevan McKinnon is a high-performance endurance coach and Director of FITTER Coaching, with 20+ years guiding athletes from age-groupers through to IRONMAN World Championship podiums. IRONMAN University Certified Coach · Triathlon NZ Level 3 · Bike NZ · NETFIT S&C. Bevan is an augo founding coach.

In this article:

  • Stress responses are not purely physical — they are shaped by emotion, environment, and life context.

  • Athletes are complex (not just complicated) systems and cannot be modelled like engines.

  • The brain acts as a central integrator: pacing and performance are regulated by perception, not physiology alone.

  • AI excels at data synthesis and pattern recognition but cannot replace trust, interpretation, and emotional regulation.

  • AI will not replace coaching — it will redefine it, making human judgment more valuable, not less.

Over the last few years, the way I think about coaching has gradually shifted. Not because I suddenly rejected data, physiology, or technology — far from it. If anything, I've become even more interested in those areas. But the deeper I went into human biology and performance science, the harder it became to see athletes as predictable machines responding neatly to perfectly prescribed training.

Why stress is never just physical

Stress responses are not mechanical reactions to physical load. They are shaped by the entire context of an athlete's life, which means identical training can produce very different adaptations depending on emotional, social, and environmental state.

A large part of that shift began with John Kiely's work on stress and allostasis. In particular, his paper A New Understanding of Stress and Implications for Our Cultural Training Paradigm challenged many of the assumptions that traditional training theory has quietly inherited for decades.

Kiely explored the idea that training adaptation and recovery are “highly individualised and heavily modulated by background emotional setting” and that physiological training and recovery are “not purely physiological phenomenon.”

That was a huge moment for me because it explained patterns I'd already been seeing for years in coaching. I had watched athletes thrive under one environment and struggle in another, respond differently to identical training, get fitter while testing worse, or break down despite what looked like “perfect” programming. I'd also seen athletes recover differently depending on life stress, confidence, relationships, or emotional state.

Kiely's work helped me understand that athletes are not simply adapting to training stress in isolation, but within an entire biological, emotional, and social environment. Once you begin to see adaptation through that lens, it becomes very difficult to return to purely reductionist coaching models.

Complicated vs. complex: why athletes don't behave like machines

Athletes are complex systems, not merely complicated ones. Complex systems adapt, self-organise, and behave differently depending on context — their outputs cannot be predicted by analysing parts in isolation, which is why training plans that work for one athlete fail for another.

Around the same time, I became increasingly interested in complexity science. Reading The Nature of Training: Complexity Science Applied to Endurance Performance by Manuel Sola Arjona pushed me deep into a distinction that changed how I saw everything: the difference between something being complicated versus truly complex.

A complicated system, like an engine, can usually be broken into parts and understood predictably. Complex systems don't behave like that. They adapt, self-organise, fluctuate, and behave differently depending on context. Their outputs cannot always be predicted by isolating and analysing individual components independently.

Humans are complex systems, and athletes are among the most complex examples imaginable. Physiology, psychology, emotion, perception, motivation, sleep, relationships, stress, energy availability, environmental conditions, and previous experiences are all interacting simultaneously, continuously shaping adaptation and behaviour. This lens fundamentally changed the way I viewed training adaptation.

When the “stress → recover → improve” model breaks down

The classic linear training model breaks down because athletes don't respond to stress in isolation. Emotional and life stress can blunt adaptation even when the physical training looks “right,” while reduced training sometimes produces unexpected improvement.

The old model was often: apply training stress → recover → improve. But real life rarely behaves so neatly. Sometimes athletes improve under less training, while at other times they stagnate despite doing everything “right.” Sometimes emotional stress crushes adaptation more than physical training ever could.

The deeper I went into this world, the more coaching started looking less like engineering and more like interpretation.

How relationships shape biology: the biosocial lens

Biosocial coaching views athletes as humans whose biology is continuously shaped by relationships, environment, communication, emotion, and social context — not as isolated physiological systems responding to training load.

That resonated deeply with what I'd experienced coaching athletes over many years. Some of the most meaningful performance changes I witnessed had little to do with interval prescription and far more to do with shifts in the athlete's broader environment.

Athletes performed differently when they felt understood, when anxiety reduced, when confidence improved, when trust deepened, or when emotional and life stress eased.

The brain as a central performance integrator

Modern neuroscience suggests the brain — not the muscles or cardiovascular system alone — regulates pacing, effort perception, and ultimately performance. It does so by continuously reading the athlete's internal and external world and adjusting output to protect the organism.

One of the most fascinating shifts in modern performance science is the growing understanding that the brain acts as a central integrator of performance. The brain is constantly reading the athlete's internal and external world — physical, emotional, contextual — and from that broader picture it regulates pacing, effort perception, and ultimately performance itself.

In other words, performance is not simply limited by muscles or cardiovascular fitness. It is regulated by a system attempting to protect the organism. That changes how we think about coaching.

If the brain is continuously interpreting the athlete's world, then coaching is no longer just about prescribing watts, pace, or intervals. Coaching also involves reducing uncertainty, helping athletes interpret effort, regulating stress, creating emotional stability, building confidence, fostering trust, and ultimately shaping perception itself. In many ways, the coach becomes part of the athlete's regulatory environment.

Where AI helps endurance coaching today

AI is already exceptional at pattern recognition, data synthesis, identifying trends, reducing admin, surfacing insights, and analysing large datasets. It will become genuinely important in endurance coaching — but its value is in augmenting human judgment, not replacing it.

All of this changed how I saw coaching. It also changed how I thought about the technology reshaping it. Like many coaches, I've spent the last few years trying to understand AI — trialling platforms, exploring new systems, testing tools, and working out where it genuinely adds value. I use it myself.

But the more I've explored complexity science, stress biology, neuroscience, and biosocial coaching, the more I've realised something important: the better we understand humans, the harder it becomes to believe coaching can be fully automated.

Because athletes do not exist purely inside data streams.

The signals that don't show up in data

The most important coaching signals are often unmeasurable. “I just don't feel like myself,” “training feels emotionally heavy,” or “I'm completing the work, but something feels off” rarely surface in heart rate variability, FTP, or readiness scores — they emerge inside trust-based human conversations.

That ambiguity matters, and often those conversations only emerge inside trust-based human relationships.

AI can analyse data extraordinarily well, but coaching is not simply data analysis. Coaching is also interpretation, communication, emotional intelligence, contextual understanding, relationship, and helping athletes navigate uncertainty. Data can often tell us what is happening, but human connection is frequently what helps explain why.

Will AI replace endurance coaches? Why it will redefine, not replace

No — AI will not replace endurance coaches. As AI takes over calculations, scheduling, analytics, training structure, and trend monitoring, the deeply human elements of coaching — empathy, interpretation, trust, communication, emotional regulation, and contextual judgment — become more valuable, not less.

Ironically, I suspect AI may not reduce the importance of coaching. It may redefine it. As AI increasingly handles calculations, scheduling, analytics, training structure, and trend monitoring, the human elements of coaching may become even more valuable: empathy, interpretation, trust, communication, emotional regulation, and contextual decision-making.

In other words, AI may automate parts of coaching while simultaneously increasing the value of deeply human coaching. Not because humans are better calculators, but because athletes are human beings before they are training systems.

The bottom line: coaching as interpretation, not prescription

Athletes don't adapt to training in isolation. They adapt within environments, relationships, emotional states, and constantly shifting perceptions of safety, stress, confidence, and meaning — which means coaching is the interpretation of a living system, not the prescription of training alone.

The journey that led me here didn't happen overnight. It evolved gradually through coaching experience, stress biology, allostasis, complexity science, neuroscience, biosocial coaching, and trying to understand where AI fits within all of it.

The conclusion I keep returning to is this: the more we understand that reality, the clearer it becomes that coaching is not simply the prescription of training — it is the interpretation of a living system.

About the author

Bevan McKinnon is a high-performance endurance coach and the Director and Owner of FITTER Coaching, working with athletes globally — from age-groupers through to professional athletes.

Across more than 20 years of coaching, Bevan has guided athletes to over 20 professional victories and titles, including national and international IRONMAN 70.3 wins, Asia-Pacific IRONMAN titles, and a top-five finish in the men's professional race at the IRONMAN World Championships in Kona. His athletes have also achieved success in elite marathon competition.

As an athlete himself, Bevan has won ITU Long Course World Championship titles, IRONMAN 70.3 World Championship age-group titles, and Kona World Championship age-group titles — experiences that continue to shape how he thinks about racing at the highest level.

Bevan is an IRONMAN University Certified Coach, Triathlon NZ Level 3 Accredited Coach, Bike NZ Coach, and a qualified NETFIT Strength and Conditioning Coach. He is also an augo founding coach, helping develop the platform in a way that aligns with his values and how he sees the future of coaching.

Where to find Bevan:

Further reading