Hey All, I am Laurens (25) from the Netherlands. Getting back into running and want to combine this with some AI knowledge

Wanted to share with everyone the framework for building your own fitness app. The build highlighted is the fun one I build live during the 4 Week Challenge video series. A little app that counts how many days in a row I've run. No code... just Claude and the gear you already wear. Here are the 7 steps: 1) Pick one idea - the app you always wanted, cut down to the one thing it does 2) Write the plan - Claude interviews you and writes the PRD it builds from 3) Design it - a flow board, a logo, an icon and a mascot 4) Build it with Claude - one screen at a time, and you test it on your phone 5) Plug in your data - the watch, ring or strap you already wear 6) Add a paywall - only if you want to charge for it 7) Ship it - TestFlight, then the App Store Come join the 4-Week Fitness App Challenge and learn to build your own! Doors close Sunday night! Kevin

Happy Monday AAC Team! We are 1,202 strong today, up from 1,166 last Monday. That is 36 new faces and we officially crossed 1,200... so freaking cool! Welcome to everyone who just joined! First, the big one. The 4-Week Fitness App Challenge is officially open inside Athlete AI Labs! Doors opened this morning: Join the 4-Week Fitness App Challenge Quick version of what it is: - 4 weeks, from your idea to a real app on your phone or the web - the app or AI agent you always wanted - Every step is a prompt you run, in order. No code background needed - Live build calls twice a week with me and everyone else building - Two 1:1 reviews with me, one on your idea and one on your plan - Lifetime access. You pay once and you have it all The first 20 seats are $249 and those 20 also get a weekly call with me every Wednesday plus my AI fitness agent. After that it's $349. Doors close Sunday night at 11:59pm ET and I honestly don't know when they open again. And same as always - the Athlete AI Community here stays free. That is not changing. Secondly, a recap of what we pushed last week in case you missed something: - 6 apps Claude can replace in a weekend - open source replacements for Strava, Whoop, MyFitnessPal, Hevy, TrainingPeaks and Zwift. One honest update: I set up SparkyFitness (the MyFitnessPal one) today to demo it and it was really weak. I would not recommend that one - I let Claude replace Strava - my own server, my own data, my own feed - 20 things you need to know to actually train with AI - The 5 biggest mistakes people make when using Claude or ChatGPT to train & how to fix them - 8 things you can do when you connect Whoop to Claude Code - On a personal note - I am pain free when running for the first time in nearly 2 months. NYC here we come!

The next big challenge in AI-powered fitness isn't collecting more data. It's making better decisions with the data we already have. Let's imagine this. You wake up after a long hike. Your legs feel like they have personally filed a complaint against you. Your smartwatch reports poor recovery, your sleep score looks disappointing, and your AI coach recommends an easy day. Then you open your training app, and another AI assistant tells you that you're ready for a personal record. 😂 Congratulations! You've officially entered the age of AI-powered confusion. The technology is getting smarter, but an important question remains: Can AI understand our physical condition well enough to give us advice we should actually trust? As of September 2026, I believe this is one of the most important challenges in sports technology. 📊 1. From Tracking Activity to Understanding the Athlete For years, fitness technology focused on collecting numbers: steps, distance, pace, heart rate, calories, and sleep duration. These measurements are useful, but they only tell part of the story. Imagine two people completing the same 10-kilometer hike. One feels fantastic the following morning. The other struggles to walk down the stairs. Same distance. Same elevation. Completely different experiences. This is where AI becomes interesting. By combining wearable data, training history, sleep patterns, subjective fatigue, and workout intensity, an AI system can begin identifying patterns that a single metric might miss. Instead of simply reporting what happened yesterday, it can help answer more useful questions: ・ 🧠 Is my current training load appropriate for me? ・ 💤 Am I recovering well enough for tomorrow's workout? ・ 🥾 How should I adjust my effort for a steep hike? ・ 📈 Which changes in my routine appear to improve my performance over time? The real opportunity isn't to create another dashboard filled with colorful graphs. It's to transform those graphs into decisions that make sense in everyday life.

After the Noop post, I went digging through GitHub this week and it turns out there's a bunch of free public repo for pretty much every fitness app we pay for... somebody already built it and gave it away. Paste the link into Claude Code and it'll install it for you. Here are the 6 I found: 1. Endurain - your own activity feed with maps, charts and friends, on a server you own. Replaces Strava. 2. NOOP - pairs with your WHOOP strap and scores recovery, strain and sleep on your phone, no membership. Replaces the WHOOP app (early Oura support too). 3. SparkyFitness - food, water, sleep and fasting tracking for the whole household. Replaces MyFitnessPal. 4. OpenGym - a full lifting log with 1,324 exercises built in, and it imports your Hevy history. Replaces Hevy. 5. GoldenCheetah - fitness, fatigue and form charts on your own computer. Replaces TrainingPeaks. 6. zoffline - ride the Zwift worlds from a server on your own computer. Replaces a Zwift subscription. (I've only set up Endurain, NOOP and OpenGym myself so far, the rest are next) Links and install steps for all 6 here: 6 GitHub Repos That Replace Your Fitness Apps Pick one, set it up this weekend and let me know which one you went with! Kevin