18th, a dodgy calf and AI in my pocket

Scott Field, T3

Eighteenth. It won’t trouble the podium photographers, but when the final Triathlon Ireland National Series rankings landed on MyTriRanking this week, 18th in my age group was the best I’ve ever finished (among eligible athletes, i.e. those who completed the minimum four national series races). It also made me the top ranked T3 athlete in the series, which I’m going to enjoy quietly!

The club asked me to write about how it happened. The honest answer is a summer that actually behaved like summer, a lot more free time than I’d planned on, and a slightly nerdy experiment with AI that kept me on start lines more often than not.

The unplanned training camp

At the start of the summer I was made redundant. It’s not on anyone’s list of training tips, but the effect was immediate. Early sessions stopped being a negotiation with the calendar. Swims got longer, rides happened in daylight on weekdays, and for the first time in years I wasn’t squeezing training into the gaps around everything else.

Then Ireland did something unusual and gave us a proper summer. I spent far more hours outside than on the wattbike, and race after race served up conditions you’d want to frame. Athy, Pikeman, Pulse and DCT were about as good as Irish racing gets: warm, calm water, kind winds, dry roads. When the weather takes away the excuses, you find out where you really are.

Claude, Garmin and an app on my phone

Now the nerdy bit. I’ve been using Claude, Anthropic’s AI assistant, connected straight to my Garmin data through what’s called an MCP connector. In practice it can read every swim, ride and run I record, and in a dedicated project space directly in the Claude interface which knows all my context and goals, this data now powers my training itinerary.

The first thing it gave me was some home truths. Back in February I was training close to twelve hours a week, and running made up just 6% of it. About 45 minutes. I was a cyclist who occasionally jogged. It also showed me that only about a third of my supposedly easy runs were actually easy by heart rate. Dublin hills will do that to you. Calf issues kept cropping up and so in May I did a full lab test (VO2 Max, Lactate, gait analysis). The results confirmed that my engine was in good shape, but my running economy was terrible. I knew this was due to a combination of weak calves and overstriding, but my adherence to the seriously boring physio prescribed exercises was very low. I much prefer going as fast as possible in the water or on a bike than being in the gym doing calf holds and raises!

So I built an app. Using Replit, I put together a small web app that lives on my phone and holds my whole plan, day by day, through to next September - including all the boring physio drills that I hate. Claude builds and adjusts the plan from my Garmin data automatically, the lab numbers and the fixed T3 sessions (Monday and Friday swims, Thursday spin, the Sunday club ride), and the app tells me what’s on today. When something changes, whether it’s a race, a niggle or a bad week, I chat to Claude with natural language and the plan gets rebuilt rather than abandoned. I’m not a software developer, so the fact that it works at all still surprises me. And the best feature? My weird obsession with ticking off to-do lists is built into the functionality of the app so I’m far more adherent to all the boring physio sessions than I ever was before.

The Summer Season

Not everything went to plan. Despite all the adherence to the physio protocols, in late August my right calf tightened up during a hard session, then again a week later on an easy run. That cost me two races I’d been looking forward to, Top of the Mourne and Skerries. Missing races when you’re fit is its own special kind of frustration.

Back to the physio. The cause is that I overstride with a low cadence, which dumps very high load straight onto my calves. I’ve been running this way all my life, and at 47 it’s so hard to change!

This is where having everything in one place helped most. The plan now has the rules built in. Easy means easy by heart rate. Tightness is the signal to stop, not pain. Single leg calf strength three times a week, with an extra set on the weaker side. The AI keeps the structure honest, but it doesn’t replace the physio, and where the physio’s programme says something different, the physio wins.

Run training was already my weak spot, and managing the calf meant my running volume this season ended up very low.

The winter project

So the off season has one job: fix the run. That means gait drills before every run, a metronome to push my cadence up a few percent at a time, heavy single leg strength and conditioning, and building run volume slowly on flat routes. Really boring stuff but it has to be done! The aim is physio sign off before Christmas, then a proper speed block in the new year.

The motivation is a big one. Next June I’m hoping to race in the European Age Group Championships in Austria for Ireland, and I’d like to turn up with a run that matches the standard of my swim and cycle. It would be great if others in the club could join me and represent Ireland in their Age Group at the European Championships - this is easily applied for on the Triathlon Ireland website and will be huge fun with sprint and standard distance race options.

What I’d pass on

• If you wear a Garmin, your data already knows more about your training than you do - it’s a gold mine!
• Easy runs should be easy. Check your heart rate, not how it feels.
• Treat tightness as the stop signal. Two missed races taught me that one.
• AI is a brilliant training partner for structure and honesty, but it isn’t a physio - ensure any workout plans are signed off by a human professional

Thanks to everyone at T3 for the swims, the spins, the Sunday miles and the company. It really is an amazing club full of amazing people. See you at the pool.