Cycling was More Fun

From My First 10k Race to an AI Training Assistant

Frankfurt Marathon. Photo: Victah Sailor.

Some years ago, when I joined a new employer, I was promptly recruited to participate in a 10k business run. My running level at the time was as good as non-existent — I did own a pair of running shoes, but they were used once or twice a year. Cycling was definitely more fun.


Without any apparent reason, I set myself a goal: to finish that 10k race within 45 minutes. The bad news: probably based on some new-colleague-wanting-to-prove-himself-bravoure I immediately announced that goal to my seriously-running co-workers, who greeted my ambitions with some wary nods. There was no way back, I was committed. The good news: I had 4 months to prepare.

Not knowing if my goal was achievable, and how to get there, I quickly went to that well-known friend of ours that has all the answers and typed “sub 45 10k” in its little search box. The results were quite fascinating:

Apparently, there are lots of experts out there with wildly varying advices, ranging from “just go out running a few days per week and you’ll make it” to detailed plans counting down every single day until the race event, even regarding nutrition and sleep.

Based on my gut feeling, I chose one of the training plans from my Google search and purchased a GPS running watch to be able to do pace-based interval exercises. I was dutifully following the training plan and quickly gathered a large pile of data about my activities, up to second-by-second measurements of heart rate and pace. With all that abundance of information, some simple questions were left unanswered by the existing applications:

  • What would be a realistic goal, given my current fitness level?
  • What should I do (today) to reach that goal?
  • How is my fitness improving?

Even when I could see my average speed increase on a weekly basis, that could be just a result of me being pushed harder by the training plan instead of me getting stronger. I ended up manually comparing detailed lap data from different activities and checking if e.g. my heart rate during certain intervals was getting lower than for similar intervals in the past. Which would take me several minutes at least, after every workout. Several applications that I tested shared the same symptoms: offering lots of details, but lacking straightforward decision support.

AI Meets Exercise Data

One of the perks of my employment at the time was that me and my colleagues were allowed to spend one Friday a month together to work on our own hobby projects and learn new technologies. Since my profession involves software engineering and data science, I quickly found ways to get hold of the raw data from my activities. One example project that emerged from our hacking Fridays was a way to visualize the technicality of a mountainbike trail based on GPS measurements:

Partial mountainbike trail with highlights for corner difficulty

Through the Strava API, some friends offered me their activities for experimentation. Analyzing these activities was fun and a great learning experience, but I always had in mind that it would be just play, and that the companies sitting on top of all that running — and cycling data would soon address my needs for better decision support. Somehow that didn’t happen.

Fast-forward to 2017

After being introduced to concepts from Lean Startup methodology, I saw a possibility of turning my hobby into something more sustainable. Various interviews and prototyping with potential users showed engagement and energy and eventually made me decide to switch from my comfy consultancy job to the uncertainties of starting a business.

The mission for FitHaxx is to provide athletes of all levels relevant and accessible support for selecting and reaching their goals. Getting more specific, this means that users will be presented with insights that are related to their goals and can be acted upon, without needing to search through charts, reports and tables full of unnecessary details. These insights will be shown in a way that is understandable for a broad range of athletes, not just the ones with a Ph.D. in exercise physiology. Finally, where possible, seamless integration with sports tracking devices will minimize the friction for trying new types of workouts.

To fully realize this mission will take some time, but as of now a first feature is available for beta testing, and you can try it out if you want! (Or read more about it here)

And How About That 10k Race?

In hindsight, the training plan that I chose for my first 10k was a bit brutal for a beginning runner like me. Thankfully, I survived it without becoming injured and the race actually went pretty well (you can view it on Strava). Since then, running turned out to be no less fun than cycling, and has become one of the main ingredients of my life. As well as turning exercise data into meaningful insights.

With xx from Fithaxx