ChatGPT built me a 12-week 10km swim training plan, designed my race nutrition strategy, calculated my kilometre splits, and coached me through every decision leading up to the Christmas 10km Ocean Swim at Coogee Beach.
Then, on race morning, I asked it one more question.
It told me to open my Maurten gels at the top and tuck them up my wetsuit sleeve for easy access mid-race. Maurten gels are basically liquid. Within the first kilometre they had leaked down both arms and into the ocean. I had 2 remaining “emergency” gels for 9kms.
That’s the thing about AI coaching. It can be genuinely brilliant. And then it can ruin your race in one bad tip.
This is my honest account of using ChatGPT as a coach — what it got right, what it got wrong, and whether I’d do it again.
Why AI Swim Coach vs Real One
After completing a few ironman events I know my body is made for endurance and for these events there is no avoiding just getting the kms in. I was naïve.
Ideally I would like to got a coach or been regularly attending swim squads training for long distance but we just had a baby a few months earlier and I couldn’t commit to consistently going during the week so had fit training around naps and supporting my wife.
That said I still wanted a challenge to focus my attention as I had Ironman New Zealand planned for March 2026 and wanted get some base built going into it.
The Coogee Christmas 10km ticked every box. Long enough to be a genuine test. Local. And hard enough to hurt. I also had a secondary goal — finish under 4 hours 15 minutes to qualify for the Port to Pub ocean swim.
I signed up solo. No paddler, no support crew. And instead of hiring a coach, I decided to build the whole programme using ChatGPT.
This was also naïve. Should of asked somebody to paddle but I wasn’t very confident I would make it to start line and didn’t want leave anybody down.
What ChatGPT Got Right: The 14-Week Training Plan
The plan ChatGPT built was genuinely solid.
Starting in mid-September and running through to race day on December 20, it structured 14 weeks of progressive swim volume built around my actual schedule — Monday squad (2km), Wednesday open water (2km from October onwards, nothing in September when conditions weren’t right), Friday squad (3km), and a Sunday long swim that built progressively from 3km up to 8.5km at peak.
It also built in a race progression strategy that I’d recommend to anyone preparing for their first long ocean swim: start with a 5km to build confidence, then step up to 10km.
The 5km at Coogee in November went well. It gave me a feel for ocean racing — the sighting, the navigation, managing effort — in a context where the stakes were lower.
Race experience beats pool volume. That’s what the plan understood. Three hours alone in a pool is useful. But nothing prepares you for open water racing like actually doing it.
Training for a 10km Ocean Swim
My training was built on volume. Four to five sessions a week, with the longest single swim being 6.5km in the pool.
I leaned heavily on fins and a pool buoy. Knowing the wetsuit would give me buoyancy on race day, I focused on building arm endurance rather than raw speed.
I was worried if this was good or bad so asked Chatgpt during my training:
Once sessions pushed past 90 minutes, I started bringing nutrition to the pool — gels or a drink — to practise fuelling while swimming.
I attended Bold Park Masters sessions on Saturday mornings. Lovely bunch of people but they had me doing Medley swimming which was good to try but not exactly helping my volume goals.
Also, some PTC technique sessions during the week – thank you Emily for helping with technique and giving me confidence I could get it done.
I was inconsistent. Work, fatigue, the usual excuses. I ended up doing most of my training solo — head down, grinding it out in the pool alone.
That was my biggest gap. Not fitness. Not technique. Accountability. Training with people pushes you harder than you push yourself. If I’d shown up to squad consistently, I’d have been faster.
If I had all time in world I would of signed up to swim smooth as their red mist sessions are legendary ☯️ Red Mist vs. CSS Training — Why Both Are Vital
The Technique Work That Actually Helped
I spent a lot of time that year working on my freestyle. Most of it came from watching Instagram coaching videos — genuinely one of the best ways I found to pick up cues and self-correct without paying for lessons every week.
I would just watch videos and try them in the pool – mainly focusing on one thing a session and try have some continuous focus on it. The biggest single improvement was my high elbow catch.
Just search freestyle swimming in Instagram and before you know it the algorithm will correct itself and its all you will see.
I was losing power on my left side because my elbow was dropping too early in the stroke. The fix was rolling the shoulder forward, delaying the catch, and keeping the elbow high through the pull phase. Simple in theory. Genuinely hard to ingrain after years of bad habits.
The other thing I worked on was hand entry. I was throwing my hands into the water too aggressively — creating lateral movement that killed my stroke efficiency. The correction: gentle hand entry, then drive the rotation from hips to shoulders. Keep the head still. Let the body rotation generate the power, not the arms flailing.
By the time race day came around, I could hold 1:50 per 100m consistently. Twelve months earlier that wasn’t possible.
Race Day: The Night Before
The swell was sitting at 1.8 to 2.4 metres the night before. The organisers had gone quiet — no updates, no reassurance, just silence.
I couldn’t sleep. Not because I was scared the event would be cancelled. Because I was scared it wouldn’t be.
This was a mistake I should have avoided. I know how to prepare for an Ironman — carb load properly, pack bags, stick to the sleep routine, treat the prep like the race. I didn’t do any of that. I lay there refreshing forecasts instead.
If I’d approached the night before a 10km ocean swim the same way I approach an Ironman, I’d have been in a far better headspace on race morning.
I always remind myself the night/day before the work has been done by reviewing my volume charts on strava.
You will notice I don’t follow the plan accordingly but more used it as a guide.
What ChatGPT Got Right: The Nutrition Strategy
The pre-race nutrition plan ChatGPT built was also good.
For a 4-hour swim at 90kg, it recommended:
- LCM bar 60–90 minutes before the start (low fibre, fast-digesting carbs)
- A Maurten Gel 160 at the 15-minute mark before the gun
- 5 gels during the swim, every 40 minutes
- Total carbs across the race: approximately 220–230g
It calculated my 1km split targets (24:00 per km for a 4-hour finish), recommended going out at 24:30–25:00 for the first two kilometres to settle HR and digestion, then holding 23:45–24:15 through the middle.
That’s a proper coaching plan. Periodised, specific, and built around my exact race context.
What ChatGPT Got Wrong: One Tip That Cost Me the Race
Here’s where it went wrong.
I write about AI. I use it every day. The training plan was good. The nutrition strategy was good. So on race morning, I asked ChatGPT one more question — how to carry Maurten gels in a wetsuit during a 10km ocean swim.
It said: open the gels at the top, tuck them up your sleeve for easy access.
The problem is that Maurten gels are extremely fluid. Not a thick gel you can fold open and reseal — liquid in a foil packet. The moment I started swimming, they leaked down both arms and into the water. Gone within the first kilometre.
I swam 9km with no fuel, no support, and no way to fix it. Solo swimmers don’t have a paddler handing them drinks. There’s no course-side nutrition. You carry what you’ve got, or you go without.
The energy drop was real. I felt it in the back half. How much time did it cost me? Hard to say. But it didn’t help.
The lesson isn’t “don’t use AI for nutrition advice.” The lesson is: never test anything for the first time on race morning. Test it in the water, in your wetsuit, at race pace — in training. No coach, AI or human, can save you from skipping that step.
The Sunscreen No One (Including ChatGPT) Told Me About
I used normal sunscreen. Not zinc. Not the waterproof swim-specific stuff.
After 4 hours and 18 minutes in the ocean, I came out looking like a tulip. Full face sunburn, and a stark white line across my forehead where the swim cap had sat. I walked up the beach thinking the white line was dirt. It wasn’t. It was the only part of my face that hadn’t burned.
Zinc-based waterproof sunscreen. Non-negotiable for ocean swimming. Add it to the checklist next to nutrition and sleep. This one’s on me — no AI was asked, no AI was blamed.
The Race: First 7km
The course was 2.5km laps. The first two went well. By halfway through the third lap I was feeling strong — making up good time on the long straight where the current was running with us.
I was quietly confident I’d crack 4:15.
The Last 700m
Then we turned around.
The final stretch — from the jetty back to the finish line — was maybe 700 metres. It took 28 minutes.
Wave after wave, relentless. Three strokes forward, pushed back. The swell had built all morning and conditions in that final section were genuinely brutal. I couldn’t see anyone around me. Just walls of water.
At one point I heard people on the jetty screaming and thought a shark had been spotted. It hadn’t. They were just cheering. I wanted to stop. Not slow down — stop.
That last section wasn’t a physical battle. My body was tired but it wasn’t quitting. It was mental. A soul-destroying, grinding feeling that the finish line wasn’t getting any closer no matter what I did.
I finished in 4:18. Three minutes outside the Port to Pub qualifier.
Christmas 10km Swim | Swim | Stravawithout the last 200m…..22:04 + 21:38 + 23:40 + 24:32 + 23:48 + 22:44 + 28:54 + 28:46 + 24:10 + 32:06 = 4 hours 13 minutes 22 seconds so technically I would of qualified only for conditions.
Missing the Qualifier By 3 Minutes
Honestly? Relief.
Missing the Port to Pub qualifier sounds like a near miss worth chasing. But qualifying would have meant trying to fit Port to Pub into a schedule already built around Ironman New Zealand. That’s a recipe for burning out and performing badly at both.
Three minutes off the qualifier was probably the best outcome for the bigger picture. One goal. One race. That’s the lesson.
The Verdict on AI Coaching
So — does it work?
For planning, yes.
The 14-week training plan was periodised, progressive, and built around my real schedule. The nutrition strategy was smart and specific. The pacing advice was sound. If I’d followed the plan consistently and shown up to squad training instead of going solo, I’d have been in better shape on race day.
For execution details tested in real conditions — no. An AI doesn’t know that Maurten gels are liquid until someone tells it they are. It can’t watch you swim. It can’t catch the mistakes you’re making in the water. That’s where a real coach, or just experience, still wins.
Free AI 12 week Swim 10km Training Plan
I took the above training pan and put it into nano banana so you can use it:
What I’d Do Differently
Use AI to plan. Use the pool, the ocean, and the squad to test everything else.
- Sort the nutrition carry system in training — not on race morning. Test it in the water, in your wetsuit, at pace.
- Show up to squad training. Solo volume has its place, but accountability is the gap. Training with people makes you faster.
- Treat the night before like an Ironman. Same prep, same nutrition, same sleep routine.
- Use zinc sunscreen. Waterproof, swim-specific, applied properly. Not the stuff from your bathroom cabinet.
Would I Do It Again?
Ask me in six months when the memory of those last 700 metres fades.
For now, one and done. There’s a flat-course 10km on the list at some point — I think sub-4 hours is realistic on a clean day with functioning nutrition.
The next big open water goal is a channel swim. Inspired partly by Ross Edgley — his approach to what the body can endure, his strength programme, his mindset. The channel is on the list. Just not this year.
FAQ
Can ChatGPT build a swim training plan?
Yes — and a genuinely useful one. It can build a periodised weekly schedule, account for races, adjust for your fitness level, and factor in real life constraints. Where it falls short is execution detail that only gets tested in the water.
How do you carry nutrition during a 10km ocean swim?
For anything under 90 minutes, you can probably get away without it. For longer swims, you need a system tested in the water — a gel flask tucked securely into your wetsuit, practised in training. Do not open gel sachets and tuck them up your sleeve. They leak. I promise.
What is the Coogee Ocean Swim?
The Christmas Ocean Swim at Coogee Beach in Perth, Western Australia, is an annual open water swimming event run as part of the WA Open Water Swimming Series. It offers 1km, 2.5km, 5km, and 10km distances.
How long does it take to swim 10km in the ocean?
Conditions make a massive difference. On a flat course, a trained recreational swimmer might target 3:30–4:00. Add 1.8–2.4m swell and a current on the final stretch, and 4:18 is a very honest effort. Give yourself more time than you think you need.
What is the Port to Pub swim?
The Port to Pub is a 19.7km open water swim from Cottesloe Beach to Rottnest Island in Western Australia. It’s one of the longest open water swims in the Southern Hemisphere. To qualify, swimmers must complete a qualifying event — like the Coogee 10km — under a set time.
Do you need a support paddler for the Coogee 10km?
The event doesn’t require one, but having a paddler carrying nutrition and giving you a reference point is a genuine advantage. Going solo is doable, but you’re on your own if your nutrition plan falls apart.
What sunscreen should you use for ocean swimming?
Zinc-based, waterproof, water-resistant for at least 4 hours. Regular sunscreen washes off. I learned this the hard way — if you’re curious what a swim-cap tan line looks like on a sunburned face, just ask.