Coachbox
Coaches were juggling spreadsheets, WhatsApp messages and scattered apps to keep track of their athletes. Training data from Garmin, Polar and other devices lived in separate silos. No central overview. No way to detect overtraining before it was too late.
I built Coachbox to change that: one platform where coaches create training plans, monitor results and communicate directly with their athletes, from recreational runners to professional cyclists and Olympians.
When an athlete trains with a smartwatch or bike computer, data syncs automatically. Coaches see completed workouts in real time and get notified if a session didn't go as planned.
Measuring training load with science
One of the hardest calls for a coach: is my athlete doing too much or too little? In collaboration with Ghent University, I developed the Coachbox Stress Score , a science-backed model that quantifies training load. Plotted on the Coachbox Fitness Trend graph, coaches can see at a glance whether an athlete is on track, at risk of overtraining, or ready for more.
Coachbox AI
Generic chatbots give generic advice. So I built an AI assistant that reasons over each athlete's real data: workouts, wellness metrics, training zones, races and annual plans. A coach can ask "compare Sarah's last two weeks" or "is she ready for a VO2 max block?" and get an answer grounded in the numbers instead of a guess.
Under the hood it works like an agent. The model calls typed tools such as
get_workouts(athlete, last) and get_metrics(athlete)
to pull exactly the data it needs, following the same Model Context Protocol (MCP) approach I use across my projects. It remembers the athlete and the coach's preferences, and it only ever sees the athletes that coach has access to, with every request logged.
Building this kind of AI tooling, MCP servers and agents is some of the work I'm proudest of. It's also why I built OpenSolo , a workspace for running and orchestrating coding agents.
What's inside
- Health monitoring with automatic sync from all major devices
- Advanced analytics , training load and fitness trends
- Smart calendar with drag-and-drop planning and templates
- Annual planning and periodization
- Workout zones with automatic calculation
- Lactate testing with threshold detection
- Strength and core alongside endurance in one calendar
- Centralized communication with athletes and team
From sports to healthcare
Coachbox is also used beyond athletics. At UZ Leuven university hospital, doctors use the platform to track the physical recovery of patients after organ transplants. What started as a sports tool now helps improve lives in clinical aftercare.
Results
- 30,000+ registered users
- 11,000,000+ logged workouts
- #2 app in the Belgian App Store (Sports category) at launch
- Scientific collaboration with Ghent University
- Clinical use at UZ Leuven university hospital
What I built
Together with two other founders, I built Coachbox from the ground up. My role: the entire technical side, from the first wireframes to the web platform, the iPhone and Android apps, and the server infrastructure that keeps it all running. Design, development and operations, all in one hand. Today I lead the technology as full-time CTO.