How my cardio AI Agent works

So a couple of months ago, I built a little fitness tracker app to record my jump rope sessions. I already made a video explaining how I built the app and how it works, but recently I've also added an AI training coach in the app, and it's probably the coolest feature in the app so far. So I want to make a quick video explaining how I built it and how it actually works under the hood. And a quick side note, I'm also working on publishing this app on the Play Store. So if you guys are actually interested in trying this out, it's coming very soon. And if anyone wants to help me out with publishing this app on the Play Store, because there's a lot of work to be done and I'd appreciate some help over here, let me know in the comments below. And now back to the video. So before we talk about the AI agent, I feel like I should give you a quick recap of what the app actually is and what it does so that you have some context, right? So this is a jump rope tracking app that I built with Claude Code to track my cardio workouts. It's got all the basic features like workout logging, history, streak tracking, goal tracking, and all the other good stuff, right? Now, the AI coach in this app is basically an agent that has access to all of your workout data, your goals, your streaks, and all the other stuff in this app. I gave it a bunch of tools to allow it to read all of your workout data, as well as change stuff like, I don't know, updating your goals or logging workouts for you and stuff like that, right? It's also the very first AI agent that I built and one that is actually useful to me. So not like some, I don't know, pet project that you build once and never use again, right? So the actual agent is built with Google Gemini because they're the only major LLM provider that I found that gave us a free tier, and everyone else is paid. And I wasn't going to use a paid solution for this app anyway. Now, I set up a backend server using FastAPI and put it on Vercel, and I used it to communicate with the Gemini APIs. And I set up the app's chat interface right in the actual application using React Native. So now users can literally just ask, you know, questions about their training in natural language and get personalized answers. Like, it's honestly amazing. Now, while this whole thing is amazing on its own already, there's another awesome feature that I built called Weekly Digest that I want to talk to you about as well. So this is basically like a newsletter or something that is automatically generated by AI using your training data and workout sessions. So all the workouts that you log in the app, right? The AI coach is going to analyze your weekly performance, see how you perform, and extract a bunch of insights from that, as well as how you can improve your training even further. It's going to write a fun little article based on all of that and give that to you. And again, this stuff gets generated automatically every week based on your training history for that particular week. And it's a really fun way to track your workout performance, see how you're doing and stuff like that, right? So yeah, that's the first actually useful AI agent that I built. And it's honestly a very fun tool to use and play around with. If you guys want to actually try this app out, like I said in the beginning of the video, so I am actually working on publishing this on the Play Store. It's just that Google's entire review process is just so time-consuming, seriously. And by the way, if one of you have already published an app on the Google Play Store before and you want to help me out with this, drop a comment below because I'd appreciate some help in this entire process, especially when it comes to the whole 12 beta testers requirement thingy, right? I'm running out of devices to install this on. Anyway, that's all for this video. Comment, subscribe, let me know what you want to see next in the comments below, and I'll see you in the next video. Thanks for watching. Bye.