Following the courses I built an AI coach that actually knows my training data. How it works: - Apple Health feeds in everything about my workouts — runs, rides, heart rate, pace, power. My watch and bike computer capture it, and it flows in automatically. - WHOOP feeds in everything about recovery — how ready my body is: HRV, resting heart rate, sleep, and a daily recovery score. - Both data streams sync into a private GitHub repo automatically, on a schedule, even when my computer is off (running in the cloud now, not just on my Mac). - I built the set of reusable "skills" — one-word commands like /recovery-check, /weekly-review, /plan-my-week, and /fuel-and-pace — so I don't have to re-explain what I want every time. Type the command, get the analysis. - The coach follows hard rules I set: never jump training volume more than ~10% in a week without explaining why, always flag overtraining or injury risk, never quietly change my profile, never overwrite past weekly reviews. And it's blunt — if I skip sessions or cut them short, it says so and ties it back to my goal race. The goal it's all pointed at: Traka 200 — a 202km gravel ultra-endurance race in Girona, Spain, May 2027. Everything — the data, the weekly plans, the fueling calculator — is built around getting me there prepared, not injured, and on pace. The dashboard: I also built a simple one-page HTML dashboard (screenshot attached) that shows days-to-race, today's readiness score, this week's planned sessions, a 7-week training volume trend, and a coaching panel with the current call on load/recovery/execution — all at a glance, no app needed, just opens in a browser. It's still evolving (next step is to explore full cloud automation so it updates itself daily without me touching my computer), but the core loop — real data in, honest coaching out — is working now. It's been a great exercise following along the course. Really enjoyed it and learned a lot. This is just the beginning and will definitely use the framework for other personal projects. Thanks a lot for making this community!

Attached is my current target fuel plan for the race. What surprises me most, or perhaps frightens me, is the 80 g carbs per hour. As a low‑carb/ keto athlete, this will require a lot of training moving forward.

Morning all. Super stoked to find this lil community! I just did the comrades marathon which is brutal 87km road run with 1,800m (80% of the climbing is in the 1st 48km) of elevation in Durban, South Africa. As a 90kg + plus guy who hasn't had a hell of alot of experience running, only 2 marathons. I decided to use ChatGPT to do some research and make a training program (attached is missing the 1st couple months when started in December 2025). Needless to say was super impressed, got my Sub 10 hour goal and felt great at the end! I work abit on tech space, and keen to start to playing around more with what AI can do here, looking now how AI has evolved the opportunities and possibilities are endless in what it can do for fitness and training in general! This forum is epic

The automation setup around day six led me down a completely different path. I no longer have a Strava subscription because I find it too expensive, and future access to API or MCP will require a subscription. After researching alternatives, I swapped it out for two tools I already had: - Apple Health — my Apple Watch already logs every run automatically, and my Garmin (which I use for cycling) syncs into Health too. So instead of Strava, Apple Health is now where all my workout data lives — pace, heart rate, power, distance, all of it. - WHOOP — I was already wearing it, so I started using it properly for recovery: HRV, resting heart rate, sleep, and my daily recovery score. This tells me how my body is actually handling the training, not just what I did. The one gap was getting the data out of Apple Health so my AI coaching setup could actually see it. For that I use an app called Health Auto Export. Currently priced at approximately usd 30 for lifetime access (compared to Strava usd 100 per year). I set it up once to automatically export my workout data on a schedule, and pointed it to save into my iCloud Drive so the files just quietly land in a folder on my phone that syncs straight to my Mac, no extra accounts or servers involved. From there, pulling the data in is simple: at the start of a session, I grab the newest export file that's landed in that iCloud folder. That gives me a fresh, readable copy of all my recent workouts, which the AI coach then reads directly to see exactly what I trained, when, and how hard. Before I fully switched over, I pulled my old Strava history one last time, 12 years of runs and rides, to calibrate my heart rate zones and spot my long-term patterns (turns out I've had a "big start, then crater" cycle for over a decade, good to finally see that in black and white). Now my whole setup runs on Apple Health for training and WHOOP for recovery. Same insight, almost zero subscription.

The dossier and logistics research is done, already ready to book the accommodation and flight tickets 10 months out!