Racefox - The director who escaped the rat race

"I've participated in the Vasaloppet several times," says Christer Norström. "And two of my kids have competed both at junior and under-18 levels in cross-country skiing." Even though Christer does not have his own competition career to look back on, this has not stopped him from developing a training tool for recreational and professional skiers. A tool that would not have been possible to build only a few years ago.

It was, perhaps, only while as a trainer for regional skiing clubs that the idea began to take hold. This role and, of course, the fact that Christer was the CEO of the Swedish research institute SICS Swedish ICT. SICS focuses on applied research into digitalisation, and its 160 researchers collectively run more than 200 diverse projects involving, among others, Internet of Things, Big Data and Automation.

"One day it dawned on me: Why don't I utilise the technology I have available through SICS? We have excellent analytic tools which should be able to be used to improve skiing. So, I invited a lot of clever people to a dinner in Kista. Among them were technicians, skiers and researchers from KTH. Then I posed a number of questions: What can be done with the technology and what kind of functionality would be like to see?"

By the end of the dinner, the group had decided to move forward and develop a training tool that could work out in the field – in an actual skiing environment. Christer and his guests were adamant that this should project should not be confined to the lab.

"We then proceeded to attach an array of sensors to a skier, to capture the skiing motion. And it was at this time that I realised that this would never be able to work."
Christer acknowledged that the idea was considered from the wrong angle. Would a 10-year old be comfortable with putting on a multitude of sensors before hitting the training slopes? Hardly. So the idea came to an abrupt halt and was shelved.

Outside on a Midsummer's Day

"I happened to hear a claim that full-body motion manifests in the torso. This idea stuck with me afterwards. So, one midsummer day, I inserted a smartphone under a heart rate monitoring belt. I then got my son to go out on his roller skis."

Christer used the sensors to measure his son's motion data. He then filtered the information and discovered that a great deal could be gleaned. When the same tests were done with a skier, it was possible to determine the skiing style employed by measuring torso movement – whether the skier was pushing off or gliding, etc. And once it was possible to determine what the skier was doing, it was not long before it became possible to analyse how the skier was doing this.

"This is when it started for real," states Christer. "First, we only logged the data and analysed it afterwards. But many people suggested we analyse the data in real-time instead. So we tried that. We learnt to identify the different motions and were able to provide direct feedback to the skier in real-time. It essentially just involves normal machine learning algorithms, what we refer to as artificial intelligence."

How to package the business proposal?

Once the technology was in place it was time for the next challenge. How could this be packaged into a profitable business?

"At this point, we realised that we had more than we originally thought. We had stored huge volumes of data, with no apparent value. But if we could draw conclusions from the vast amounts of data, we would be able to link this to simple tests to determine how good a skier is. Thereafter, we could incorporate this into the Vasaloppet. Then we would be able to determine what was needed to complete the Vasaloppet in a specific time, which technique should be practised to reach this and at which pace this should be done."

No sooner said than done. Christer Norström and his colleagues collated Vasaloppet data from a multitude of skiers pertaining to performance and motion data from the skiers to be trained.

"Let's say you're a race three skier. We know your strengths and weaknesses and which goals you have set for yourself. By comparing the skiers in our database we can determine that you need to practise to achieve greater upper-body strength. Based on this, you are assigned an exercise regime and your progress is continuously measured so that you can achieve your goal. You get instant feedback, just as if your coach was right there with you on the track, encouraging you and providing you with new, perfectly attuned exercises. This is how we packaged our tool – as a real-time coach."

Trial and error before upscaling

The coach, naturally, does not only have to focus on skiing. The tool works for various sports, but Christer – and the company which was originally named Wememove but has now become Racefox – is not particularly well-positioned for this yet.

"We'd prefer to go all the way with skiing before taking this further. Of course, we've also developed prototypes for running, because it is a global phenomenon. But before we focus on this we must first have a well-established Go To Market strategy. And it's good for us that skiing is a smaller sport, internationally speaking. It enables us to test things in relative peace."

A somewhat surprising future may also exist in equestrian sports. The Racefox tool can, for example, be used to determine whether a horse pushes off on all four legs or if one is slightly weaker. An experienced sulky driver can also see this, but Racefox is able to detect early signs of limps. This means that the primary market may not be in helping drivers win races, but in preventing injuries to horses. But this remains a vision for the future. Right now, skiing is the main focus.

"When we become ready to scale up our operations, we'll be able to do so quite easily," says Christer. Technically, it's not actually difficult to provide for 10 million users or even more. All we need to do is scale up memory and computing power through the cloud."

Hardware has become low tech

Everything has become so much simpler and cheaper owing to the cloud. A decade ago, Racefox was forced to use an array of servers for data storage. They had to build devices to provide feedback to users. Today, everyone has a smartphone. So they had to establish a sales network instead. Nowadays, everything can be sold via Amazon.

"A crucial factor is that hardware has become low tech," states Christer. Whenever you need something it can be purchased very cheaply. For the most part, you can create a new smartphone brand in three months at incredibly low costs. In contrast, it has never been more difficult to make money."

Expect a few million!

Neither Christer nor his colleagues at Racefox had any extensive experience with hardware. Yet, they still opted to develop the hardware – a heart rate band with integrated motion sensors – on their own at a low cost. Once they figured out how to do it, that is.

"We started by contacting a few Swedish companies to develop a prototype. They quoted us around SEK 800,000. Then they said we could expect it to cost a few million more before production start-up."

Somewhat disheartened, the Racefox team began searching for alternatives.

"After a few simple searches on Alibaba, we found several suppliers who had pulse belts utilising the communication protocols we required. We contacted them and received our first response within 45 minutes."

The future belongs to startups

The suppliers were able to make the changes required by Racefox and developed three functional prototypes at a cost of around USD 5,000, of which USD 2,500 needed to be paid upfront. The prototypes arrived in the post six weeks later. After a few iterations, the supplier developed a low-cost mould and Racefox was able to place its first order for 1,500 units.

"The cool thing is that it wasn't more expensive or difficult than that! It's hard not to speculate about the future. When looking back to industrialism and the time following World War II, it was the age of colossal companies. But I believe that the future belongs to startups. The combination of connected devices, cloud networks, AI and open data is unbelievably powerful."

It's safe to say that Christer Norström does not regret the journey on which he and his colleagues have set forth.

"I've worked as a leader in the industry for many years and, prior to that, as a professor and vice-rector within academia before starting at SICS around seven years ago. During those years, I would never have believed that running my own company could be so much fun! And, above all, I'm glad to have escaped the rat race. It's so easy to be complacent with a good salary and never take a chance on something else."