18:57
In September I ran a 5km time trial in 21 minutes and 16 seconds. I decided this was unacceptable, set a target of sub-19, and built a training plan.
The plan had heart rate zones, pace targets, aerobic decoupling thresholds, and weekly volume caps. I built a custom web app called Runistry to analyse my Garmin data because Garmin’s own analytics weren’t granular enough. I used Claude — an AI — as my coach, feeding it my data after every session and getting adjusted training prescriptions back. I tracked cardiac drift percentages to two decimal places.
I did all of this to run 5 kilometres slightly faster than before.
By mid-December I was flying. Two consecutive weeks above 60km. My threshold pace was dropping. My VO2max intervals were coming in at the fast end of every target range. I told my AI coach I was ready. The data agreed.
On January 11th, after six weeks of preparation, I lined up at 6:40am for the time trial. I ran 3.47km, decided I didn’t want to do this anymore, and stopped. I walked home. In my training log I wrote “DNF due to mental quit, not physical limitation,” which is a clinical way of saying I bottled it.
Two weeks later, at 6:11 on a Saturday morning, I tried again. Nobody was watching. My wife didn’t know I was attempting it. My kids were asleep. My parents would find out later via my cousins, who found out from my uncle, who saw it on Strava. Which is a very 2026 way to learn your son did something.
The run was a slog. I wanted to quit before kilometre one was done. I wanted to quit again at two. And again somewhere in the third. Each time I had the same negotiation: I am not stopping again. I would rather run 19:20 than stop at 18:40.
I was behind my splits. A few seconds per kilometre, which sounds like nothing until you’re doing the maths in your head while your lungs are burning and you realise you need to find fifteen seconds somewhere in the next eight minutes.
Kilometre five — slightly uphill — I ran the fastest split of the day. I don’t know where it came from. Maybe the cool morning air. Maybe spite.
18:57. Done.
Here is what I find genuinely funny about all of this.
I had an AI coach that adjusted my training plan in real time. I had a custom-built analytics app because the one made by a billion-dollar wearable company wasn’t good enough for me. I had a spreadsheet that calculated aerobic decoupling to two decimal places — a metric I couldn’t define without Googling it six months ago. I drank electrolytes prophylactically, like a man preparing for surgery.
The thing that actually mattered was deciding not to stop.
All that infrastructure, and the critical variable was stubbornness. I could have achieved the same result with a pair of shoes and a bad attitude.
In 2021, I ran 17:39. Five months of training, boom, done. I posted about it on LinkedIn with hashtags. I told people to challenge themselves with a single ambitious goal. I used the phrase “crystal clear.” It was, in retrospect, the kind of post that makes you want to throw your phone into a canal.
Four years later, the goal was a minute and twenty seconds slower. It took more effort. I failed the first attempt. And I told nobody I was trying. I think this is called growth. Or maybe ageing. Probably both.
It’s been a week now. The satisfaction lasted about three days. I’ve started looking at Hyrox events, which involve rowing machines and sled pushes and are designed to make people suffer in a different configuration.
I don’t fully understand why I do this. I didn’t need a sub-19 five kilometre time. Nobody was asking for one. It has no professional value. It won’t make me live longer — though not doing it, I’m fairly certain, would make me die sooner. Not from cardiovascular decline. From the slow rot of having nothing difficult to do.
Maybe that’s the whole thing. You need a hard thing. Not because the thing matters, but because you without a hard thing is a worse version of you. The AI coach, the analytics app, the training plan — that’s just how my brain justifies the suffering to itself.
18:57. Nobody saw it. It mattered anyway.