I have just come out of the gym after my first proper session in a while. Last week was a family trip to London, which was brilliant, but it knocked me out of my routine.
This morning was one of those days where I really did not fancy going at all. WHOOP had my recovery in the yellow. Sleep was fine, stress and strain yesterday were nothing dramatic, but I still felt flat. It would have been very easy to decide that today was not a gym day and leave it at that.
Instead, I tried the new approach I have been talking about. I took my WHOOP numbers and dropped them into ChatGPT. In return, I got a clear session plan with target figures that matched how my body was supposed to feel on a “medium” day.
The structure was simple. Five minutes of warm up at a set pace to get moving. Then three blocks of 2,000 metres on the rower, again at a set pace. Nothing heroic. Just long, steady, repeatable work.
On paper it looked almost too easy, especially with that “you should probably train” yellow score. In reality it was exactly what I needed. Each 2,000 metres felt long and cruisy. Hard enough that I knew I was doing something, nowhere near the point of blowing up. By the end of the third block I felt like I had trained, but I did not feel broken.
The bigger difference was in my head. I walked into the gym tired and not really in the mood. I walked out feeling lighter and quietly pleased with myself. The combination of WHOOP data and ChatGPT as coach gave me just enough structure to get over the hump of not wanting to start.
It is early days for this experiment, but right now it has promise. If this is what a “didn’t want to go” day can look like, I am curious to see what happens on the days when I actually feel ready.
It feels like I have a new coach. You might remember that I’ve been asking ChatGPT to prepare workouts and coaching advice to help me row 2,000 metres in under seven minutes. That is the whole point of The Sub-7 Experiment. Can I use AI to help me achieve a fitness goal.
When I first tried this, the model available to me was GPT 4.0 and it was good. I set up the role, gave it the context, kept the conversation alive, and each day I told it how I was feeling. It responded with the workout for that day. That simple rhythm worked well.
Then 4.1 came along. The improvements were small but noticeable. It held context better, got confused less often, and could handle my slightly lazy and vague questions more easily. As with most tech, each release pushed things forward.
Earlier this year GPT 5.0 arrived with a huge amount of hype. It was supposed to push ChatGPT into a whole new league. PhD-level reasoning. Better accuracy. A genuine step up.
The reality was mixed. Yes, it produced better code for a different project I was working on. Yes, it had more “thinking” capacity and could reason more deeply without me having to explicitly tell it to think. That part was a massive improvement.
But as a conversationalist it was a step backwards. It forgot things. It lost context. It got confused about tasks. The backlash was so strong that OpenAI reinstated version 4.1 as a choice because so many people preferred to keep using it. It felt like my coach had left the building.
I worked around it with careful prompting, but it was frustrating. The Sub-7 Experiment relies on continuity and rhythm, and something was always slipping.
Which is why it now feels like I have a new coach. GPT 5.1 has arrived and it feels different in a very good way. The tone is consistent. The help it offers is actually useful. It anticipates the next step instead of fumbling it. It feels like an upgrade in the true sense of the word.
Let me explain why this matters.
I have a WHOOP device and I’ve used it for almost two years. Before ChatGPT became part of my training, WHOOP was my only guide. It sits on my wrist and picks up all sorts of measurements: heart rate, skin temperature, strain, sleep quality and plenty more. Every morning the app shows three dials. Yesterday’s strain, today’s recovery level, and last night’s sleep. When you start a workout you tell WHOOP what you’re doing and it gives you a target strain. When you hit it, the band vibrates and tells you to stop. Simple. Clever. And it worked well.
But in the last few months I’ve been questioning the value. The subscription model has changed. The promise of free device upgrades for active subscribers has been replaced by an “uplift fee”. My renewal is in February and it will cost a lot more to keep going into my third year. And that makes me ask what I actually need it for.
Most of the data WHOOP collects I don’t really use. I know how well I slept because I was there. What I actually value is recovery guidance and strain targets. And there are other devices out there that do similar things for a simple one-off cost. Polar Loop is one I’m looking at seriously.
So I turned to ChatGPT 5.1 for help. Reviews. Recommendations. Thoughts based on my training. And one of the threads pointed out something obvious: I have not been using most of WHOOP’s data anyway. Not deeply. And the only thing that truly matters is the workout planning, which comes from CoachGPT.
I asked if there was any way for ChatGPT to access WHOOP data directly. It said no, the APIs are not available yet. But then it made the suggestion that genuinely impressed me.
It told me exactly which two screens in the WHOOP app to screenshot each morning. It told me to upload them, and it would analyse everything it needed: recovery, sleep, strain, HRV, and readiness. It would then produce a fully tailored workout for that day. And if it thinks I need a rest day, it will tell me that too.
I have used this new process for the last few days and it is genuinely brilliant. Two screenshots. Upload. Instant plan. Clear reasoning. Exactly what it expects from me. Exactly what to avoid. Exactly how hard to push.
And it works. The coaching is better. The structure is better. The whole system feels like something new.
Well done, ChatGPT 5.1. This is The Sub-7 Experiment. Recovering using structured data.
Tuesday, and it’s been a good couple of days. Yesterday I saw the doctor and got the all-clear after both procedures, signed off, back to normal life, and very happy about it.
So today I went to the gym, mostly to reconnect with the rowing machine and see where I’m at. Knocked out 7,000 metres in 31 minutes, lower intensity, low drag factor, but smooth and steady. The old me would’ve jumped straight back in and pushed too hard, too soon. This time, I’m not doing that.
Coach ChatGPT’s advice is still in my head: take it easy, rebuild properly. I might’ve gone a little harder than planned, but it felt great. The legs worked, the form felt good, and more importantly, the head was clear.
So, we’re climbing back into the fast lane, not quite there yet, but picking up speed. And that’s enough for today.
This is The Sub-7 Experiment — recovery mode disengaged.
It’s been a while since I’ve written one of these. The last big thing was the 150km cycle — a brilliant day out. But not long after, I came down with a cold. Headachy, wiped out, all the usual. And just as that started to ease, I went straight into leg surgery two Mondays ago.
Today was my first time back in the gym in weeks. Literally weeks. I’m nowhere near full power. I did some rowing (even though I’m not really supposed to) and a few light weights — nothing heavy, just enough to move, to feel like I’m doing something again.
And I do feel weak. Underpowered. Off. But I know this isn’t a new baseline. This is a blip. A temporary stop on a longer road. I’m grateful more than anything — grateful that I’m on the mend, that I have a gym to go to, and that I can move at all.
I’ve got another surgery in two weeks, so I know I’m not fully back yet. The sub-7 experiment is still very much alive, but realistically, it’s not going to happen this yar. And that’s okay.
What I need to hold onto now is patience — something I’ve never been great at, especially with myself. But recovery takes time. Push too hard, too soon, and I’ll just end up further back. So this is where I am: moving, mending, grateful, and trying to let time do its job.
This is The Sub-7 Experiment. Slow for now, but still moving.
Just back from a tough but rewarding ride; 75km out and back, with 1,220 meters of climbing. No loops, no shortcuts, just there and back, with a headwind all the way home. It was tougher than I remembered: the kind of ride that demands your full attention, your legs, and most of your patience.
What made a real difference this time was the fueling. I spent yesterday making banana flapjacks and cheesy eggy rice cakes (shout-out to ChatGPT for the recipes and fuelling strategy). They worked. I didn’t feel nearly as wrecked as I normally do on a ride like this. I still had something left at the end. I probably could’ve used more water, but I stopped about two-thirds through yo top up my bottles and to see me home.
The legs are well and truly cooked now, but in that good, earned way. Possibly not helped by Thursday’s 135kg leg presses, probably not the smartest prep, but lesson learned.
This was my longest ride of the year, and while it wasn’t easy, I’m pleased with it. There’s satisfaction in pushing through, seeing that distance logged, and knowing you got it done.
So that’s today’s entry in the Sub-7 Experiment—which, as I keep discovering, is about far more than rowing. It’s about effort. Growth. Trying. Failing. Learning. And showing up again tomorrow.
Not sure what’s up today—this is the third time I’ve tried to do this. Maybe it’s me, maybe it’s the app, maybe I just keep hitting the wrong button. Whatever. Let’s call this one off the cuff, because that’s exactly how it feels.
The important bit: another good session.
Strong. Organised. Full-on. Bit of rowing, bit of weights. Finished with a couple of fast 250m sprints, let the engine rip, felt solid.
Keys went in the box again. That’s becoming a habit now. Quiet statement, every time: I’m here. I belong.
It’s Friday. The weekend starts now. That’s all I’ve got. Just a check-in. Job done.
The Sub-7 Experiment 