I Used ChatGPT to Train for an Ironman Race and It Helped My Time

Triathlon is an pricey hobby that I stumbled into about two years ago when a friend asked if I’d sign up for an Ironman with her.

The $700 race entry fee was only the beginning of the expenses that ensued. I had to buy a bike, a helmet, and cycling shoes; a swimsuit, goggles, and a wetsuit. All in, I spent more than $5,000 to get across the finish line of my first Ironman race.

My second go-round at the 140.6-mile distance would be less expensive, as I’d already invested in the necessary equipment, but it still wouldn’t be cheap. And I didn’t make things any easier on my wallet by traveling abroad — to Kalmar, Sweden — for the race.

In an effort to curb my expenses even just a little, I opted out of paying for a training plan through a site like TrainingPeaks, which can cost around $100, or hiring a coach, which could run hundreds of dollars a month.

Instead, I turned to ChatGPT.

In late April, about four months before the race, I provided the AI chatbot as many details as possible — when the race was (August 16), my goal (to better my time from two years ago), my strengths (running), my weaknesses (cycling and swimming), and how much time I could dedicate to training (15 to 20 hours a week) — and asked it to put together a plan for me.

Within seconds, it started spitting out information, including a weekly layout, periodization blocks, and sample workouts.

Here’s the weekly schedule it came up with.

It also provided a few “key rules for success,” emphasizing that, for me, “bike volume is king” and to “prioritize your long rides no matter what.” Plus, it gave a few tips for the weeks I flagged when I’d be traveling: “Drop all training on flight days — pure physical and mental rest.”

I requested some adjustments to the weekly layout. I noticed it only prescribed two runs, plus one optional recovery run. I’m used to running every day of the week, so I asked if we could tack on one more weekly run. I also asked if we should incorporate a run workout so I wouldn’t lose speed.

It replied enthusiastically: “Excellent question — and yes, 100% you should,” and listed the many benefits of speed workouts.

Good thing I flagged it.

I didn’t mind the back-and-forth exchange. It felt like we were working together on a group project for a subject I love, where studying feels more like play than work. I felt qualified to push back and make my own suggestions. But if this were my first triathlon, or I didn’t already do independent reading on the subject in my own time, I’d want someone qualified to tell me exactly what to do, and trust that I wouldn’t have to suggest modifications to their program.

It didn’t feel like a student-coach relationship — we were peers — and it was telling me everything I wanted to hear.

Consider this exchange.

KE: What are your thoughts on stretching? Personally, I hate doing it.

ChatGPT: You’re definitely not alone — a lot of serious endurance athletes feel the same way about stretching. Let’s be real: stretching is not magical — and hating it doesn’t mean you’re hurting yourself.

Whether it’s right about stretching not being magical or not, it’s appeasing me rather than challenging me. A real, human coach — a good one, at least — will push and challenge just enough to get the most out of their pupil.

Another observation I made as I went back and forth with ChatGPT was the random errors it made. They were small and often related to the date. For example, this sample weekly schedule started on Monday, May 13, but May 13, 2025, was a Tuesday, and May 14 was a Wednesday.

This type of error happened a few times and, while not catastrophic in this scenario by any means, it surprised me that even AI doesn’t get all of the details right. I made a mental note to fact-check, especially when using ChatGPT for more important tasks than creating Ironman training plans.

I spent about an hour finalizing the program with my AI coach and, shortly after, started implementing it.

The training itself was a lot of hours, but felt relatively relaxed and pressure-free. Having followed a strict online plan for my first Ironman, it often felt like the training dictated my life. I’d receive an email every morning detailing the workouts of the day, and when I completed them, I’d receive a color-coded “workout card.” Green meant I nailed it, while orange meant I was way off, and red marked an incompletion.

This time, I shuffled workouts around to accommodate travel or plans with friends guilt-free, without worrying about a poor report card.

I wasn’t worried about the lack of accountability because I genuinely like most of the training. I didn’t feel like I needed the extra motivation that a coach or training plan can provide to get out there and do the work. All I wanted was a general weekly structure to follow and some workout ideas, which AI provided.

Objectively, the plan worked: I bettered my time by nearly two and a half hours.

When I went back to my chat history to share the results with Coach GPT, it wrote: “You Did the Hardest Thing: Not just crossing the finish line — but doing the daily work for months, even when it wasn’t exciting. That’s Ironman. That’s resilience. That’s awesome.”

Those are nice words, but I imagine they don’t come close to the feeling of picking up the phone, ringing your coach, and delivering the good news.