Today we're going to talk about why long rides might not be the best thing to prepare you for long races. We're also going to talk about a big mistake that you could be doing with your training that could be affecting some of your key workouts. We're also going to talk about swimming, possibly being a secret hack for cycling and maybe why, Nate, I did a race over the weekend and I might be disqualified. I just found out just before recording this. Is you litter? No, I did not. Everything was in my pockets. Yeah, yeah, we'll cover that. We're going to get into that. You're very much against that. Yeah. Oh, yeah, totally. I hate it. I've stopped. Yeah. 100% in the middle of a race. I've dropped a gel and I stop and go pick it up. I do not. I don't tolerate. So it wasn't literally. I don't drop it. I don't stop in the middle of a race, I've dropped a gel and I stop and go pick it up. I do not. I don't tolerate it.

I don't drop it.

I don't stop in the middle of a race for sure. Yeah. If it's a feed zone, I won't stop. But otherwise. Yeah. But there's another reason. But before we do all that, Nate, it's good to have you back on the podcast. We should probably jump in. give some people some updates, you and the team.

Everyone's been head down, working hard, really hard. on the podcast. We should probably jump in, give some people some updates, you and the team.

Everyone's been head down, working hard, really hard on tons of stuff.

Yeah, it's been, the world has changed. And with all the new, like, I feel like when we

launched Traynard AI, the AI stuff was the models were just getting along. And now there's

like so many months have passed. And it's, you know, 10x is good. It's crazy. The tooling and everything. And I haven't on the podcast a lot, but it's like 15 hour days every day. I'm listening with my kids. It's crazy. You see like the amount of commits that I have. It's like double anybody else.

We track it in the company.

Yeah, we track it in the companies.

We can see commits of all the engineers and stuff.

And Nate's top in charts.

They were, but we've doing these AI meetings and everyone else is like catching up.

It's going to, the gap's going to close.

So anyways, train art AI. It's been going well the gap's going to close. So anyways, uh, trainer at AI. It's been going well,

except for a few people who are, uh, I want to talk about some of the things that have,

happened and what we're working on and what has been fixed, what hasn't fixed. So right when we first

launched, there were some things with like, um, we need switch a workout and the workouts like

cascade down and they change. Uh, there were some bugs around that on our code side. And when I talk about this, there's two things. There's code issues where it's like, those are deterministic things where it's like one plus one equals two. We can look at it. We can fix it. And then it comes up right away. Then there's models side. And that's with our ML models. And these are our own models. We're not using LLMs. Like these are things that we've trained. It takes about 12 hours to train our model right now, depending on things that might get it faster. But we want to make an update. It takes 12 hours with the result. And that's the ML thing. It's kind of a black box. You don't really know you have to build like test harnesses around it.

See what's happening.

In general, if you do indoor workouts, you will get, like, you'll get faster.

It works really well.

...