It’s one of the hot topics of recent times, artificial intelligence. It’s met with equal parts excitement and cynicism from different groups, with some claiming it will revolutionise our lives, and others decrying it as an affront to everything that is decent. The answer, as expected is probably somewhere between the two.
The first thing to acknowledge is that there are different types of artificial intelligence out there. ChatGPT is where most people’s minds go to when they think of AI, but this is a relative crude, blunt application of the technology. It churns out poor quality images and text that is for the most part, easily identifiable as being created by AI. In its current form it is certainly no threat to triathlon coaches, as plans that people have created using ChatGPT prompts have been riddled with errors, or otherwise ridiculous jumps in volume/intensity. Suffice to say, we should not feel threatened by this level of artificial intelligence replacing coaches.
However, there is more to AI than ChatGPT. Programmes which are built from the ground by triathletes for triathletes which leverage artificial intelligence. TriDot, who I have recently signed up to as a coach, is an AI based triathlon platform that has been around since 2011, but has recently been thrust into the spotlight with as other AI applications have become more advanced.
It works by taking information on your thresholds and training history to write you a plan that helps you build towards races in your calendar. It then uses information on your execution of sessions to adjust the plan on the fly. If you miss a session, participate in a group training session, go over distance or otherwise go off plan, the AI will re-write the rest of the training to accommodate for this.
For me, the obvious benefit is to those who work shifts that change at short notice, very unpredictable jobs or otherwise struggle to get all their training in for reasons out of their control. Although as a coach I am very happy to re-schedule somebody’s training around their evolving week, often these people are too busy and tired to message me to ask me to rearrange, and simply cram in what they can, when they can. Instead of an interval session in the pool they do a 30 minute run, and instead of a strength session they hop on the turbo because it’s all they have time for. This isn’t inherently their fault, but it is frustrating for me as a coach to see my carefully written plan go up in flames for reasons beyond control for either of us, and feel unsure how best to help this athlete achieve their goals.
However, if they were using TriDot, the AI would recognise that they had missed a bunch of important sessions, and adjust on the fly. So, does this mean that I am out of a job as a coach?
Not quite, the problem is that at the end of the day, AI is still a blunt instrument. It’s still a computer making decisions based entirely on data. AI doesn’t think, it doesn’t feel, it doesn’t appreciate context, it simply churns out numbers.
One of the misconceptions surrounding coaching is that the training plan is the most critical component of training. That when you engage a coach you are paying for their training plan that will help you achieve your goal. This has given rise to a coaching model where a coach at a local tri club who went to Kona once runs a business selling “personalised” training plans to club members. They charge an arm and a leg for the same plan they give everyone else, with minimal ongoing support. Struggling athletes are told to suck it up, and can even pay extra for the privilege of being shouted at from poolside once a week.
However, in my opinion the training plan is probably the least important component of the coaching package. Yes it is of course important, but when you go on a high end coaching course there is very little taught in the way of programme design, it is much more focused on the relationship between a coach and the athlete.
No matter how advanced a machine is, it can’t call up an athlete and offer them words of reassurance when they’re unwell and panicking about lost fitness. It can’t read the signs that they might be on the verge of overtraining, help them get their tech setup, recommend how to become more aerodynamic, set them a non generic strength programme, advise on race day fuelling, provide recommendation on how to manage niggles, know the kind of training an athlete responds best to, provide insights on supporting their training with diet, and provide support when things inevitably don’t quite go the way we planned. Even if within our lifetime AI evolves to perform these kind of tasks, it will only ever be cold and clinical. While there may be a minority of isolationists who want an emotionless machine to tell them how it is, the vast majority of us want a compassionate human who has been through the same experience themselves and knows the right thing to say at the right time. Physical preparation is only part of the triathlon puzzle, and this means AI will only ever get you so far.
The issue with all AI is that it is based on its existing dataset. TriDot runs an annual research project where someone gets two months of training in exchange for letting their training data influence their AI. By increasing the data available to TriDot, they improve the accuracy of the training.
However, this is also a large flaw. How many really successful athletes are going to fire their coaches to try two months of AI training? there may be a handful, but if an athlete has had great success with one methodology, it doesn’t make sense to throw caution to the wind and totally upend their training approach for a couple of months free training. The result of this is that TriDot will have limited training data from certain groups of athletes. There are no doubt those out there who have used it to qualify and compete at Ironman World Championship events, but if you are playing around with the idea of turning pro or training for an event type not supported by TriDot, then this isn’t the best option for you as the. data pool probably just isn’t large enough.
Equally, if you train as part of a group regularly then TriDot is not for you, and they admit this much in their marketing. If you have a club swim on Monday, run session on Wednesday, long run on Saturday and group ride on Sunday, the artificial intelligence is going to tie itself up in knots and spit out some truly garbled training recommendations.
The sessions prescribed by AI are also quite repetitive in many ways. While triathlon training requires quite a bit of repetition as there are certain sessions you just need to do on repeat, if you are an athlete who struggles with their attention span, you will likely be better off working with a coach directly or self coaching if the social and fun aspect is more important to you than performance.
If you know you have an interruption to your training such as a holiday, it can also be very difficult to help the AI understand this. You can just run and lift weights in the hotel gym when away, but you’ll still get emails every morning telling you to do a turbo session or pool swim which is frustrating and makes the feeling you’re falling behind on your training while away all the more palpable.
The cost is also something to consider. While TriDot starts at £15 per month, you can pay up to £129 per month for the full raft of features. For £120 a month I could write you a truly bespoke plan written from the ground up around you, with a monthly call where you can update me with how the next month looks for you and your availability, we can discuss your diet, plan your race calendar together, I can be there to reassure you if the doubts creep in, and provide all of the services AI could never hope to replicate. Although the plan won’t auto adjust on a day to day basis, if one month is a total disaster I can take this into consideration when writing the next month of training. These are factors to consider when deciding on the best approach for you.
Artificial intelligence really shines when it comes to streamlining repetitive tasks, or performing tasks that we as humans could never hope to. It saves coaches a lot of time, freeing us up to spend more time with athletes delivering meaningful coaching, instead of spending several hours a week writing plans, which even if you’re a top level coach, requires a lot of copying and pasting workouts to make small edits to them week to week.
There are also additional tools available in TriDot that are unavailable elsewhere. It uses a more modern version of TSS, adapts training sessions based on the environment you’ll be training in, can base your training around your genetic profile (if you have one), help create a race plan, analyse your swim stroke and more.
As referenced above, AI is especially useful for those who would struggle to stick to a rigid training plan. If you know you will be walking out of work at 5PM on the nose every evening and not be expected to be in again until 9, don’t have a family of many other commitments you may be able to have great success with a rigid training plan for a fraction of the cost, but most of us have dynamic, unpredictable lives which need to be accounted for.
It’s also the best option for those who simply don’t want a coach, or can’t afford one. For many it’s just not affordable or desirable to hire a coach. Personally I’m not sure I would be very coachable given my knowledge of the sport, so I have recently started using one of the cheaper TriDot packages to set my training. I will report back after a few months, but so far I have found it to work quite well for me.
Yes, as ever the middle ground is most likely a sensible option, which is where coaching with AI comes in. By combining the advanced features that AI can provide with the common sense and experience of a coach, you really can reap the rewards of both.
You have the real time edits to your plan that only AI can provide, as well as a coach who can spot when the artificial intelligence has got it wrong, and intervene. When you pick up a niggle running, can’t work out why your stomach is so unhappy while running, need someone in your corner when you’re full of doubt and just need someone else who understands triathlon in your life, a coach is there to provide what an algorithm could never achieve, a human connection.
While £129 for TriDot’s premium offering is pretty punchy, for an extra £70 you can work with a coach. In my eyes, this is a complete no brainer, in fact it’s an absolute steal. A coach could well be the difference between you achieving your goal or not, so paying a small premium to work with one is one of the best deals out there. If you can afford to spend £129 a month on your training, I’m almost certain you could stretch to £199. Carbon fibre wheel sets and oversized jockey wheels offer a minimal return for a significant investment, where this coaching offer is some of the best money you can spend in triathlon.
Artificial intelligence will not replace triathlon coaches, however it will probably put bad coaches out of business, and have a negative impact on the sale of static training plans. However, this is simply part of advancement in technology and the evolution of a craft/business. As a coach who wants to be in the game for another 20+ years, I cannot afford to bury my head in the sand and tell myself it’s a passing fad. I need to understand and embrace it if I still want to be in business in five years time.
However, I am still not completely up to speed on ins and outs of it all. The ink is still drying on my coaching agreement with TriDot, and I need to learn the platform, so I have an offer to make you.
My standard rate as a TriDot coach is £299 per month, however I don’t feel comfortable charging this kind of money before I know the platform well.
The first three athletes (article published on 12/1/25) who contact me via Simon@phazontriathlon.com can, assuming that our introductory call goes well, work with me for £199 a month for the first year of training. That’s £100 a month in savings.
If that is out of your budget, you can use TriDot from £15 per month, and I even have an exclusive link to give you a two week free trial: https://app.tridot.com/onboard/sign-up/simonolney
There are those out there who will trip over themselves to use AI to support their training, and others who wouldn’t touch it with a barge pole, but it’s here to stay, and as a professional in the coaching space I’m choosing to embrace it, rather than go the way of the music executive who was adamant that Spotify was a passing fad and people would go back to CDs.