How I Built an AI-Driven
Application for Endurance Athletes
Hi! Who are you and what business did you start?
Hi, I’m Markus Rummel and I’m the founder of AI Endurance. AI Endurance is an AI-driven software application that provides personalized guidance on how to reach performance goals for endurance athletes, i.e. amateur runners, cyclists, triathletes.
What is your personal story and how did you come up with the idea?
I’m a passionate triathlete, runner and cyclist with a PhD in theoretical physics. I have 10 years of experience in scientific research, including artificial intelligence and machine learning. With a full-time job and family, training time for my hobby is limited. To get the best possible results in races, I’ve always been looking for ways to train efficiently and optimize my training.
As endurance athletes we collect tons of data such as running speeds, gradients, heart rate, cycling power etc that influence performance. However, the apps I was using in my training were only telling me basic information about my past activities – my best 5k time, the number of kilometers I cycled in a given week, etc.
I wanted an app to tell me how to train optimally based on all the data I’m constantly accumulating. With my work in physics on machine learning I knew that there would have to be a way to accomplish this and utilize the full potential of that data. This is exactly the mission of AI Endurance and why I created it.
A crucial aspect of AI Endurance is personalization. Everyone responds differently to different training regimes. I see it with my friends all the time! Some can hammer it every day and their performance grows whereas others need to take a day off more often than not to improve.
AI Endurance is a tool that lets every athlete find their own unique path to improvement and success. Endurance sports can be so much more enjoyable when the initial struggle pays off. You become faster and perform beyond your expectations at your goal event date.
What challenges did you face when creating your product/service?
Coming from a purely academic background there was – and still is – a lot to learn every day to create a successful product. Developing a backend that runs a machine learning algorithm is only a small part of it. A successful product also needs a great user experience, a well communicated value proposition, and a creative and effective marketing campaign to name a few.
Another big challenge continues to be the limitations around race events due to Covid-19. The cancellation of all race events resulted in significantly less demand for an app that brings you in peak shape on a given race date. AI Endurance has pivoted with a ‘Maintenance Mode’ option. Users can maintain their current fitness level until they can get into more serious training when races do come back.
It’s also a challenge in terms of marketing, finding users and feedback. We were counting on representing AI Endurance at many local races.
Who is your target market?
The target market is ambitious amateur endurance athletes (runners, cyclists, triathletes).
How do you market your business and which approaches have been the most successful?
Most of our users were acquired via unpaid efforts such as word-of-mouth, social media and SEO.
Before races were cancelled due to COVID-19 in March 2020, we also had some success with piggybacking on race events via vlogging, boosted blog posts on Facebook, and low competition Google Ads.
Since you launched, what has worked in not only attracting but retaining customers?
We really try to honour our customers, offering help and going out of our way wherever possible and maintaining regular and personal contact, mostly via email. We were happy to have early adaptors reach out to us when we initially launched the beta version and provide us useful feedback which influenced our current version.
More recently, paying customers, who arguably have a higher interest in the product working for them, have reached out as well. We appreciate all user feedback and take it all into consideration.
This strategy has led to numerous improvements of the app. In our situation where we have this powerful AI in the background from which we can infer different kinds of information, we really rely on our users to tell us what they find most helpful in their training.
What kind of culture exists in your company, and how did you establish it?
We value a diverse and inclusive culture at AI Endurance. A cultural aspect I’ve adapted from the academic institutions I worked for is the results-focused, hands-off approach. No individual is judged by their background, gender, race, etc and also when and how they work or how much or how long.
It’s only the results that are important. We motivate each other with positivity in our communication which encourages self-motivated individuals. We are a research-active company, currently writing a research paper on machine learning with one of our summer interns.
What software, services or tools do you use within your business?
AI Endurance’s stack is a React frontend and a Django backend, hosted on Google Cloud. For the first prototype we used a Django app hosted on Pythonanywhere which was lean and quick to implement. Continuous Integration (GitHub, Docker, Kubernetes) allows us to implement changes and fix bugs quickly.
One thing I can recommend to other founders is to track when you’re doing what – I use Clockify. It allows you to reflect if you’re spending your time wisely. It keeps you sane when you feel like you haven’t accomplished anything if it’s one of those days where you’re just putting fires out.
What are the most important lessons have you learned on your business journey?
If I would start over again I would make more of an effort in the earliest stages to get my idea in front of potential customers. I definitely underestimated how hard it is to even make people try your product out for free.
From the development perspective, I feel like I have a much better sense now when to put the time in to create a ‘proper’ solution and when to implement a ‘band aid’ solution. It’s a tough one when time and resources are so limited.
You want to create something scalable and robust but often it’s a better decision to first gauge the market. This saves time and effort that can be used more wisely on something that people actually want.
What is your favourite aspect of being an entrepreneur?
Building something new. To come up with your own solutions when there are no blueprints and have something actually working that didn’t exist before is an amazing feeling. It’s actually not so dissimilar from solving an original problem in science and requires much of the same skills such as creativity in problem solving and persistence.
Constantly learning new skills and meeting many new people from all kinds of different backgrounds are also great aspects of being an entrepreneur.
What is your LEAST favourite aspect of being an entrepreneur?
It can be overwhelming. Sometimes it feels that with having to do so many things at once – nothing will ever get done. That’s why it’s important to prioritize and focus on the things that will really bring the business forward. And take a deep breath, every long journey starts with a single step.
What books, podcasts or other resources have inspired and influenced your business journey?
For me, being part of the local Hamilton (Canada) startup community has been the single most useful resource for AI Endurance. A huge shoutout goes to Innovation Factory and the startup incubator The Forge who have mentored me on the business aspects of running a startup. Bouncing ideas off of and learning from other successful entrepreneurs was incredibly helpful.
Where do you see your business 2-3 years from now?
A big long-term goal is to grow into a leading automated scientific training platform in the sports and exercise space. AI Endurance’s approach to predict performance based on individual training data can be readily extended to other areas. For instance, prediction of general fitness, injury prevention and weight loss from past training data offer a huge market opportunity for the future. As performance tracking via the internet-of-things increases, more opportunities in this growing market will arise.