Mission

Our mission is to automate staying healthy by combining the psychology of habit building, the analytical capabilities of machine learning, real-time health data from sensors, and the latest in preventative healthcare research.

Team

Through our journey, we’ve built a team of likeminded passionate people, that came together around a shared purpose of making healthy living easier and more approachable to all people, not just those who already know what to do.

In our previous digital lives we’ve worked for companies like Google, Bayer, Tigerspike, Deloitte, Unilever, and Rolls-Royce.

Origin story

Healthzilla started as an app to automate workouts with intelligent algorithms to calculate sets, weights, and reps, and has evolved into our vision of automated wellness. Where did we start and why? How did we get to where we are today?

 

The idea for automating health started with a personal problem for our CTO Aki. As a young computer geek and engineering student, he went through mandatory military training in Finland as was introduced to the art and science of strength training. What ensued was a lifelong interest in the quantification and analytics of health. Initially this took the form of spreadsheets to try and take the art out of training, and make it something quantifiable and trackable through numbers. For years, this took the form of spreadsheets and printouts carried to the gym for testing. Yes, this is before the iPhone.

The formal genesis for the idea to create an app was the release of the Swift programming language by Apple. It promised to make hardcore app engineering more approachable, and that sparked a challenge in Aki to spend nights and weekends hacking together an initial prototype.

Initial concept work for the first app prototype. The goal was to implement the most robust scientific strength training programs such as Daily Undulating Periodization, and have the app calculate and optimize your weights and reps for you while managing progressive overload.

Initial concept work for the first app prototype. The goal was to implement the most robust scientific strength training programs such as Daily Undulating Periodization, and have the app calculate and optimize your weights and reps for you while managing progressive overload.

Having totally immersed himself in the prototype for almost a year, Aki suggested to Laura the idea of starting a company around the concept. We engaged external designers and developers to rebuild the app into something that could be published on the App Store. Laura jumped in as full-time CEO and we incorporated the company in 2016.

We expanded the coverage of the programs to include cardio, mobility, and bodyweight exercises. We shot videos to include instructions for each exercise. We created fitness tests that would quantify the user’s baseline within a program personalized for their goal: longevity, weight loss, strength, or cardio. The result of the fitness tests would design a weekly schedule of training appropriate to the user’s level, and each program also contained rules for progressing based on the results from each workout.

We still think it’s the most sophisticated workout app on the App Store!

Going from sketches to wireframes to designs was an experience. Despite the low number of screens, the complexity of the underlying content, programs, and algorithms make it a very complex build.

Going from sketches to wireframes to designs was an experience. Despite the low number of screens, the complexity of the underlying content, programs, and algorithms make it a very complex build.

One year later we finally had a product out, and quickly realized a fundamental limitation of the concept. We could help users automate some of their workouts, but…

a) Fundamentally the fitness market is driven by marketing and trend-following fads, everyone just wants the easy silver bullet even if it never works, because the real solution is too much hard work.

b) We still couldn’t get couch potatoes to start working out just by offering an app to them. That would limit the reach and mission of our work dramatically. We wanted to make change happen, not just facilitate the fit becoming fitter.

Our post-launch analytics started telling a story of user behavior and also lack of habit change. People who already went to the gym continued to go to the gym. Habit change is hard!

Our post-launch analytics started telling a story of user behavior and also lack of habit change. People who already went to the gym continued to go to the gym. Habit change is hard!

Before investing too heavily into the app we had just spent a lot of blood, sweat, and tears building and launching, we started over. Well, not really over, but we felt we hadn’t solved the problem. We wanted to reach higher and address a wider audience.

Enter the dragon. At this point, we also rebranded the app to Healthzilla, introducing the synonymous fiery friend to make the boring theme of health data and analytics less serious and more approachable. We didn’t want this to feel like a visit to the dentists office!

Initial wireframes for the new design of our second major release, introducing the character of Zilla and broadening the scope and ambition of our feature set and target users.

Initial wireframes for the new design of our second major release, introducing the character of Zilla and broadening the scope and ambition of our feature set and target users.

With the surprising global success of the Apple Watch, wearables were here to stay. New devices were coming in at every price point, and sensor technology was improving with every generation. We had gone from steps data to real-time heart rate and sleep data in just a few years.

We saw this as an opportunity to use all this valuable raw data as insights for people to understand their health, and apply this data in improving their health through exercise. The recently announced CoreML framework from Apple finally allowed Machine Learning models to be run on the device. In combination with Apple’s on-device health database HealthKit, this meant we could both access and analyze the data without having to upload anything to the cloud.

The availability of interesting physiological and behavioral data in combination with on-device machine learning inspired us to develop algorithms to predict the users stress levels based on trend analysis and pattern matching. We could also use the same approach to offer the user population level benchmarks to answer regular questions like “is my resting heart rate normal for my age?”.

Some of our early models to recommend fitness goals to users based on their age and body-mass-index. Users classified as overweight for their age would be recommended weight loss programs, while those skinnier than average would be encouraged to build up their strength at the gym.

Some of our early models to recommend fitness goals to users based on their age and body-mass-index. Users classified as overweight for their age would be recommended weight loss programs, while those skinnier than average would be encouraged to build up their strength at the gym.

After a few months of building now with an in-house team, we were able to get Healthzilla out into the market to gather more feedback and data. What we found, surprisingly, was that still many users did not have wearable devices despite the growth. There is now way to filter out such users in the App Store, and we wanted to offer something to users that had no data to analyze.

So we put our heads together once again and started researching plethysmography, which is a method of capturing heart-rate data from an optical sensor. In English, that means using the camera on your phone to read a (very weak) heart rate signal from the capillary veins in your fingertip. More so, we wanted to use that signal to extract the extremely valuable yet sensitive heart rate variability data. This is your body’s #1 KPI for stress and recovery.

This is the kind of signal you can get from an optical sensor like the camera. Notice the different waveforms that need to be filtered and void of artefacts to produce an accurate enough signal to extract heart-rate-variability.

This is the kind of signal you can get from an optical sensor like the camera. Notice the different waveforms that need to be filtered and void of artefacts to produce an accurate enough signal to extract heart-rate-variability.

We were excited to make a free resting heart rate (“RHR”) and heart rate variability (“HRV”) scanner available to the market for free, to enable anyone with a phone to start measuring this data. Now available on iOS and Android, just click the app store links in the footer below!

Are we done here? Not by a long shot… stay tuned.