ABOUT · WHY RAYVIK EXISTS

The plan said 4:15. The clock said 5:30.

I've been in tech my whole career. Long hours, always at a desk, always on a screen. The kind of job where every day feels like putting out fires.

Running was my way out of that. I started eight years ago, mostly just to keep my body going. But it stuck, and eventually turned into halves, then fulls.

The moment it really hit me was my first full marathon.

The afternoon that settled it

Going in, I had three halves behind me, best time around 1:47. I gave ChatGPT that number and asked it to plan my race. It ran the conversion formulas, the same ones anyone can google, and told me confidently that sub-4:30 was no problem at all. So I set my targets: stretch goal sub-4:00, main goal 4:15, floor 4:30. The safe number.

Race day was hot. Nothing dramatic, just that flat, ordinary heat that grinds you down. I hit 26km and felt my legs start to go. Stopped at a portaloo. Came out and they were done, completely locked up. Every few steps a cramp would fire. I couldn't run anymore. I walked and shuffled the rest of the way.

Final time: 5:30. An hour past my floor. I crossed the line just after one.

FIRST MARATHON · THE RECEIPT
Half PB, going in1:47
ChatGPT's prediction, from that halfsub 4:30
Legs gone26 km
Actual finish5:30

The post-mortem wasn't complicated. I'd trained like a half-marathon runner. I had done a 35km long run, but the overall mileage just wasn't there. The strength work wasn't there either. And ChatGPT had no way of knowing any of that. It just saw a 1:47half and drew a straight line. It sounded so certain. That's the part that stings. Nobody who has actually cramped at 26km would have given me that answer with that much confidence.

FIRST MARATHON · THE LINE IT DREW, AND THE LINE I RAN
5:30 · me+1:004:30 · chatgpt026 km42.2 km
Dashed: the conversion formula, extrapolated from one half marathon. Solid: me. I was ahead of it at 26 km and an hour behind it at the finish.

What the chatbots kept getting wrong

So I went looking for something better. I tried using ChatGPT and Claude more seriously for training. And honestly they impressed me at first. But the longer I used them, the more I realized they had no idea who I actually was. Every session, I'd be pasting in screenshots of my pace charts, my sleep data, my weekly mileage. A few minutes of data entry before I could even ask a question. When HealthKit integration finally arrived, I thought okay, now we're getting somewhere. But the advice didn't really change. Still generic. Still felt like reading a textbook that happened to have my name in it. And they kept forgetting things. I'd tell them about a tight calf, and two weeks later they'd have me doing intervals on it. I'd build a whole race plan with one of them, and a month on, it was like that conversation never happened. I was the one carrying all the context. I was the memory.

That's not a coach. That's just a very expensive notes app.

I went looking for something that already did this

I tried Garmin myself. During a run, I understood the appeal: heart rate was easy to check at a glance, and spoken workout cues were genuinely useful. Before and after a run, though, both the watch and the phone app were painful to use. It felt like driving a Tesla whose touchscreen ran Windows 98: capable hardware, with simple tasks buried behind software that fought you at every turn. And after all that data, I still didn't get a clear answer to the question I cared about: should I train today, and how hard?

Then there was WHOOP. It knows your body well. But the AI felt slow and passive. Strength training was almost comical. You had to follow its templated formats, and if you wanted something specific, you'd end up creating the whole plan yourself. A real coach would just tell you: three sets of hip thrusts, Bulgarian splits, squats, plank. Maybe send you a quick video. No jargon, no setup, just a conversation. Something you could talk to like a friend on WhatsApp, and it already knows everything about you.

Nothing I found came close to that. So I started building it myself.

What I want at 6 a.m.

Not much, honestly. I want to wake up and know today's call. Rest, easy run, intervals, or the gym. That one decision, made well, every morning, for months.

Making it well takes the last two months though, not just last night. The niggle in my right leg. The race in eleven weeks. Sleep, HRV, and the three days I'll spend in a hotel gym next week. If I have to assemble all of that by hand every time, collect it, paste it, explain it again, the friction wins. I skip it and I guess, and I guess in the direction I already wanted to go.

THE CONTEXT BEHIND ONE DECISION
last night · a screenshot06:12 · today8 weeks backright ITBhotel gymrace · 11 wks
Left of today already happened. Right of it is already scheduled. Last night is the black sliver — one day of the nineteen weeks the call actually depends on.

So the answer should be sitting there when I wake up. Already thought through. Already aware of what happened yesterday and what's coming on Saturday. It knows my history, my injuries, my race goals, my sleep last night, the fact that I'm traveling next week. I built that for myself first. Then enough friends turned out to want the same thing, and it became a product.

A plan is not a PDF

The other thing 26km taught me is that a fixed plan is a fiction the moment life touches it. Work runs long. Family needs you. You get sent to another city for a week and the long run has to move. Most products hand you a table and hope you march through it, and the first time you can't, you're not off-plan, you're just off.

Bodies differ too. Some people are built for distance and some aren't, and no conversion table on the internet knows which one you are. It doesn't know your race is at altitude, on a rolling course, in heat, either. Those things get planned for in advance, or they get discovered at 26km.

So the plan has to move. Hold the goal and move everything else. Reallocate the week when the week collapses. Add work when you're absorbing it, and call a halt when you're not. Something that keeps rewriting itself toward the same finish line, instead of a schedule you fail against.

The part where I ignore my own coach

A few months after I started using it myself, Rayvik flagged me. HRV down, sleep down, recent load too high. It told me to back off the long run. I felt fine, genuinely good actually, so I ignored it. Ran it hard. Did the same thing the week after. Nothing happened right away, which is always the trap. Then my right ITB started aching. Then one morning I got halfway through a run and realized I was running a fever.

SIX WEEKS · WHAT IT KEPT TELLING ME
ran it hardagainHRV51 ms↓ 6 weeksSLEEP6h 06↓ 6 weeksLOAD67↑ 6 weeksright ITB
Two things falling, one climbing, and the two long runs I ran hard anyway. The ITB turned up after the second one.

Eight years of running. I built the thing. And I still didn't listen.

That's exactly the coach I wanted to make. One that remembers everything, sees the full picture, and is willing to say no, especially on the mornings when you're convinced you don't need it.

If you ask most runners why they started, the answer is almost never a finish time. It's usually something quieter. Wanting to feel better. Wanting to be stronger. Wanting a bit of time that's just yours. The PB stuff comes later, and somewhere along the way it quietly takes over. You start trading your body for numbers, and calling it progress.

I don't think that's what training should be. A good finish time should be the byproduct of actually getting stronger over months, not something you borrow against your health to buy on race day.

That's what Rayvik is built around, and it doesn't much matter where you start from. Whether you've never run before and just want to get moving, or you've got years and finish lines behind you, underneath it's the same thing. You want your body to be better than it is. You can't sit still.

Wherever you're headed, HYROX, a first 5km, or a marathon that owes you one, we want to get you there steadily.

Get stronger, steadily. That's all we're trying to do.

Thanks for reading this far. I'd guess you've got your own small plan to be a bit better than you are, and your own reason for it. If Rayvik can help, come .

— Kai, founder of Rayvik