Cycling Ruined My Sleep—AI Showed Me Why

There’s a simple recipe to turn frustration into new skills on and off the bike.

AI Is a Game-Changer for Cyclists

3, 2, 1. AI is a game-changer for cyclists like us. I just solved a huge cycling-related issue that I’ve been dealing with for years in minutes. Don’t worry if you’re not on the AI train yet; that’s okay, you’re not alone, I assure you. But after today, it may just be your new best friend. And just as I was about to hit record on this, out comes Garmin’s new AI, Garmin Connect Plus.

The Timing Is Perfect

The timing is perfect because if you haven’t been neck-deep in ChatGPT, Grok, or Gemini, it’s going to be really hard to judge if Garmin’s new subscription is worth the 10 bucks a month. We’ve tried them all—free versions, paid, you name it. Glen’s building apps that would have taken months in an afternoon, like it’s crazy.

Solving My Cycling Problem

And I have some thoughts about Garmin, but what I want to focus on first is how I fixed a difficult fitness problem with AI so that you can see for yourself how powerful it is. Let me tell you, when I shared this story in my weekly email, I had the biggest response yet—people with the exact same issue as me. Here’s how I describe AI for fitness: Imagine you go to see your doctor, and you arrive, and she’s got all the time in the world for you.

She’s brushed up on the latest peer-reviewed studies—like all of them. She’s also asking about things like your VO2 max, that’s new, how you’re fueling your rides, and your recovery habits. She wants to see your sleep data, and she has a question about your bike fit. She assures you that together, you’re going to drill down until you’ve pinpointed your unique issue. Weird, right? Well, that’s AI.

AI Helps You Solve Problems

And while I’m in no way suggesting you replace your doctor, there are issues that you’re going to solve on your own, and for the more serious ones, you’re going to arrive at your appointment well-armed. Because in reality, at our age, as cyclists or runners, we’re not normal. We’re outliers. Most people our age are not fit like us, and so our issues can become a bit more exotic.

AI Spots Patterns

Until now, because now we have AI and their large language models, or LLMs, that are trained on massive amounts of data, so it can spot patterns. Unlike a Google search that just spits out links based on keywords, making you do all the work and hoping you find something relevant, AI reasons through massive amounts of data it’s trained on, connecting dots and spotting trends and giving you answers tailored to your specific situation.

Search just matches words; AI understands context, and it will crunch things like your Apple or Garmin data to find solutions no one else is suggesting, as I’ve discovered. So today, I’ve got five ideas to help kickstart your AI journey in fitness, gear, tech, setup. I’ve even got some ideas for Garmin. But let’s start with my own freaky problem so that you can see how I solved it, because honestly, I didn’t think I was going to.

The Freaky Problem

I mean, who else could get up nice and early, put her bike on the roof of the car, mountain bike for hours in the sun in the fresh air, and then not sleep that night? I mean, what does it actually take? So even my Zone 2 rides this winter were causing me to wake up at, like, 3 or 4, like, for the day. It was brutal. And to really appreciate how amazing AI is, I need to give you a little bit of context.

I’ve Tried Everything

When I tell you that I have implemented every expert tip to facilitate better sleep over the years, I really mean it. It is my priority because, without it, as you know, everything else—our rides, diet, motivation—is so much harder. I reordered my life around better sleep. I stopped drinking alcohol, I do the morning sunrise walk every day, I dim the lights after sunset, I go to bed and get up at the same time every single day, and, of course, I exercise.

And we’re eating dinner so freaking early, I feel like, “Well, what time is it? It’s 4:30. 4:30? Who eats dinner at 4:30?” By the time you sit down, it’ll be 4:45. We gotta catch the early bird; it’s only between 4:30 and 6:00. Yeah, they give you a tenderloin, a salad, and a baked potato for $4.95. You know what that costs you after 6? If we eat at a decent hour, I’ll treat, okay? And to be clear, those changes have transformed my sleep, except for the days I ride, which is four times a week.

What AI Told Me

What are you going to do? You going to stop riding your bike? And it dawned on me: I use Grok and ChatGPT for research all day, every day. So I popped over to Grok and said, “I do four 1-hour rides in the winter; three of those are in Zone 2. I’m finding my sleep isn’t great on those nights; I wake up at 4.

Why is that?” I got a stellar overview of steady-state rides in Zone 2, including this bit: While it’s fantastic for endurance and overall health, it can subtly influence your nervous system and hormones in ways that might affect sleep. And then it suggested five sleep-disrupting culprits, and immediately my eyes locked on number three: Depending on your fueling, or lack thereof, your blood sugar could dip overnight.

If you’re not eating enough carbs or protein post-ride, your body might release cortisol or adrenaline to mobilize energy, jolting you awake at 4:00 a.m. I was like, “Jolting awake at 4:00 a.m., sometimes 3?” Okay, so at this point, I’m getting excited. I tell Grok about a typical post-ride low-carb lunch and provided some sleep score data, and Grok immediately starts analyzing things like protein timing, recovery load, and circadian rhythm.

So I tell it about all my lunches and my breakfast and when I work out, and it tells me, “This is gold; your details are locking it in.” I love that. So I tell Grok I’ve been low-carb since the year 2000 and that I don’t even sleep after three-hour mountain bike rides, and it declares this a breakthrough moment: “Low carbs since 2000, dialed-in discipline, and now we’re cracking why sleep’s dodging you despite epic rides.” And then Grok figures it out.

It explains, “You’re dipping into reserves overnight, nudging cortisol up and waking you up at 4:00 a.m. Alertness on waking says it’s not exhaustion; it’s energy regulation.” Overnight, your body’s like, “Where’s the energy?” So it’s all happening so fast, and just like that, Grok suggests a test: I’d add a third of a cup of sweet potato—yeah, that specific—to my low-carb lunch and see how I sleep that night. Easy.

And guess what? I slept like a freaking baby. But you know, I know one night does not make a trend. So after each ride, I’d go back to our chat and I’d say, “I just did a Norwegian 4×4 ride; I’m having chicken salad for lunch; what do I need to add?” And check out what I get: energy burn from my ride, a breakdown of my lunch ingredients, and the carbs I need to add.

And then one good sleep turned into a week and now a month. And not to brag, but this just happened: I pitched a no-hitter. And Grok figured out that mountain bike season is just the same issue times 10. I’m going all day without fueling, and I was just in a big old glycogen deficit. So without having to ask, it went ahead and created a summer preview for fueling.

It solved a huge problem for me, and it turns out I’m not a freak. When I shared this in my weekly email, I started hearing back from subscribers with the same issue—someone on a keto diet, another with type 2 diabetes who had resorted to sleeping pills, both stoked to figure it out with AI. And by the way, if you’d like to get our weekly email, you can sign up right here.

Five AI Ideas for Cyclists

Okay, enough about me, because this is really about you. Here are five ideas to get you started with AI. And after that, I’m going to give you my thoughts on Garmin Connect Plus and whether you should pony up for it, plus an idea for Garmin that I think is long overdue. So let’s begin with the obvious: optimizing your training. So think about a specific goal you have.

Maybe you’ve got a road trip in Mallorca coming up, or your first mountain bike trip in the Rockies, or you need to improve your VO2 max, like I talked about in my last video. Did you know that you could cut and paste screenshots of your data right into ChatGPT or Grok for analysis? Here’s a screenshot from intervals.icu that I pasted into Grok.

It extracted the data, did a comparison between rides for power and heart rate, analyzed the data, and gave me a recommendation from a picture in seconds. Like, that’s wild. You could share your Garmin training status and ask questions; it’s hyper-personalized. And what about nutrition? A lot of us get that wrong. We’re either eating too much on the bike, like I talked about here, or not at all, like me.

You might say, “I’m 57, I’m riding 80 miles tomorrow with 1500 meters of elevation; what should I eat?” The grams of carbs per hour will be tailored to your age and intensity, along with hydration tips. Then afterwards, go back and let your AI know how it goes so it can adapt over time because it’s learning about you. And better nutrition helps us to balance rides with recovery so we can avoid overtraining this season.

As I noted in another video, on Garmin’s website, they say, “Train too hard, too often, and you may never get back to your peak fitness.” Garmin’s recovery hours and training status can be frustratingly opaque, so your prompt might be something like, “I rode 70 miles yesterday; Garmin says my recovery is 54 hours, my HRV is 45 milliseconds today; should I ride or rest?” If you’re sitting it out today, it might give you time to sort out your gear and tech issues.

Gone are the days where you have to Google action camera reviews to scan random opinions. Now your AI will figure out whether an upgrade is worth it, like something Glenn’s considering for our rides. It asked Glenn a few questions about what he values in an action camera and did some deep research to compare features, performance, and feedback from users.

My friend Richard, who has a 2013 mountain bike with a weird seat tube, found four compatible dropper posts in seconds with the free version of ChatGPT. Maybe it’ll help you sort through the myriad of gravel bike choices or suggest a firmware update or switching to whatever GLONASS mode is for better satellite reception on your Garmin Edge 530.

And speaking of maps, you might want to use it to help plan a route or for trip planning. That’s what I’m doing right now. I’m thinking about a cycling trip in northern Thailand, and if anybody has done that and you have advice for me, can you let me know in the comments? ChatGPT is weighing in with the collective wisdom on things like how many nights should I spend in Bangkok.

I dropped 15-page itineraries from tour companies to create a list of experiences that I might want to replicate on my own. It could also help you plan your cycling routes for trips so that you balance hard days with easier days while experiencing the best of a destination. And that’s just five easy ideas using some of the best tools out there, like Grok, ChatGPT, and Gemini, and all of them have free versions.

Garmin Connect Plus: My Thoughts

I wanted to expose you to what’s possible before I weigh in on Garmin’s new AI because, honestly, I don’t want your first exposure to AI to be this, because your reaction is going to be, “Uh, what’s all the fuss about?” You guys know I love my Garmin data; it’s just that their new subscription model isn’t much more than that. Here’s their five big selling points.

Active Intelligence, arguably the most important in my opinion: these are insights and recommendations on sleep and activities throughout the day. Cool, except that the intelligence is stuff I already know. Like, after my weightlifting session yesterday, I just couldn’t wait to see my new insights. Here they are: “You completed a strength training session today.

” Yeah, I’m aware. And then it tells me the data that I’ve already got in the app. And this life-changer: “You’ve taken 4,080 steps today; your goal for today is 6,920 steps, which means that you still have 2,840 steps to reach your target.” Math. When I go into the HRV widget, that’s where I love some personalized insights, but it’s just the same stock messaging that you’re getting right now.

Moving along to the Performance Dashboard: you’re now able to create graphs and overlay data, like, say, heart rate and power over your zones for the month. And that’s really good; it’s just that I think we should have just had that ability in the past as part of what we get with the Garmin ecosystem. So to me, this really isn’t AI; it’s reporting.

And it’s why I connected my Garmin to intervals.icu for free. Would you pay for this? I want to know; let me know in the comments what you think. So the next three features, I’m going to just skip over very quickly. Live Track allows others to track you during an activity. I’ve never used that in the past. The new enhanced Live Track lets you send it in a text notification, not just an email, to friends or family when you kick off an activity to your own page.

Okay, is that worth 10 US a month, or that’s Canadian, 10 Canadian a month? Training Guidance says you’ll get expert guidance for Garmin Coach plans, and I did do their Gran Fondo plan a couple of winters ago, and I really liked it. And based on what I’ve seen so far, I think you’ll be getting what I call richer content, like, say, some strength training videos in there.

But my understanding is that the underlying Garmin algorithms really haven’t changed. But you know, I really did save the big feature for last: badges and a star, kids. Exclusive badges. Look, I don’t want to turn this into some kind of flex for those of you that are not Garmin Connect Plus, but I’ve got the star for about 30 days.

Apparently, they double your Garmin points too, which, until yesterday, I didn’t know existed, and personally, I don’t care about those. But here’s what I really do care about and where Garmin should be going. Recently, I’ve been working with Grok on my Garmin data, which I’ve managed to royally screw up, and this is on me, not on Garmin. I didn’t realize that my Garmin power pedals needed to be calibrated each ride.

It’s super simple; I just didn’t do it, and I polluted my data for months. So my watts were hugely inflated, and on top of that, I was lazy and I wasn’t wearing my heart rate strap. So now that everything’s tickety-boo on that front, what Garmin sees is a history of, like, big-ass power and low heart rate. So even though I’m actually getting fitter, Garmin thinks all my recent rides are strained or unproductive, and my VO2 max is inflated.

So I just don’t think that I could be the only one whose inputs are wrong for a while. I think there are probably tons of people with wrong data, and I suspect some don’t even know it. So what are we supposed to do, like, just delete our data and start fresh? I want to be able to have a conversation with Garmin and reason through my problem, like I showed you.

I think if we could say something like, “Hey Garmin, my power and heart rate data wasn’t right from this date to this date; however, my rides from this date to this date are correct. Can you go and please adjust my data and training status and estimated VO2 max based on that?” Grok says that’s totally within the realm of AI. At the very least, we should be able to say, “Please disregard my cycling activity data prior to February 13, 2025, and base your guidance on activities from Feb 13th forward.

” Like, when you think about it, Garmin’s AI should have alerted me to the fact that, unless I just qualified for the Tour, something’s likely off with my data, and it should have suggested to me what to look into to make sure I’m not polluting my data and things I care about, like VO2 max. That’s intelligence; that’s what AI is all about, and that’s what I’d pay for.

Like, Garmin’s sitting on a mountain of data; its intelligence and guidance should be next level.

And maybe in time it will, but that time is not now. I’ve got the 30-day trial in my calendar so I’ll remember to cancel. I’m going to continue to use Grok and ChatGPT to crunch my data until they figure it out. And until then, I’ll continue to enjoy what I think they do best, because I love my Fenix watch and my free Garmin data.

In fact, I’ve changed some really big things in my life based on it. I talked about that in this video, and it’s starting right now. See you soon. [Music] [Music]