Tonal Engineering Onsite and Hackathon!

Tonal Engineers and Data Scientists connect in a week long onsite @HQ

The team doesn’t just love fitness, they love art too!

For a week in June, Tonal’s Engineering and Data Science Teams congregated at Tonal HQ for a week of innovation, brainstorming, and connection.

Tonal has team members all over the world! It’s great to get together in San Francisco and enjoy a thrilling, zesty round of Mini Golf. But our onsite wasn’t just fun and games, we also used the opportunity to come together and tackle complex technical challenges, including:

• Developing engineering-led initiatives for this next year

• Diving into challenges we’re facing and rework our processes to address them

• Presenting at the Databricks Data+AI Summit

• Two Day Hackathon

Hackathon

Everyone’s favorite part of the onsite (except the Mini Golf) is our hackathon. We spent two days developing prototypes we'd love to see incorporated into Tonal. This year, we developed 15 projects! From AI to engineering overhauls (server driven UI anyone?) to user facing features, our team explored it all.

This year the competition was tight. So tight that the judges couldn’t decide between two projects, so they both won.

Hackathon Winner 1: Tailored Recommendations

For winners Jacob and Sungsu, personalization is a key pillar at Tonal. They saw an opportunity the program and workout recommendations on the Tonal’s home screen. While some rows are personalized to a member’s goal, they wanted to dive deep into providing tailored recommendations for our users. They explored two ways of deepening this functionality:

1. Introducing Contextual Content Rows

Users are often looking for a certain type of workout. These new recommendation rows are based on key points that members might consider when choosing their next workout, including:

• A harder workout for a member’s primary goal (progression)

• An easier workout for a member’s primary goal (regression)

• A recovery workout

• A novel stimulus or modality

• Workouts that target underutilized muscles

These rows help members better understand what to expect. This additional clarity also promotes greater autonomy, allowing members to choose the path that best aligns with how they’re feeling on a given day.

2. Recommendations Based on Training History

Once a member selects the most relevant path, the second part of the project personalizes the workouts within that row by leveraging their recent training data.

For both progression and regression rows, we use our content scoring system alongside a member’s training history to identify the level of workouts they've recently completed. We then apply up to a 30% adjustment—either upward for progression or downward for regression. This ensures we’re tuning exposure based on what each member is prepared for. We also surface this progression or regression contextually, so members can gauge how each workout compares to their recent activity.

For the novel stimulus or modality row, we assess the member’s typical workout patterns and recommend less-familiar formats—like yoga or Pilates—for days when they might want something different.

In the case of underutilized muscles, we take a historical view of training distribution across body regions (Lower, Upper, Core) and opposing muscle groups (e.g., Chest↔Back, Biceps↔Triceps, Quads↔Hamstrings). If we detect an imbalance, we recommend workouts that target the undertrained areas.

Conclusion

Ultimately, this hack-day project empowers members by offering clearly defined workout paths tailored to their preferences and readiness. It personalizes not just the selection of content, but also the contextual information around each option—making it easier for members to make informed, confident choices about their fitness journey

Hackathon Winner 2: Remote Control the Tonal

This project, built by super team (Ivan, Sampath, Adam, Noah and Baha), lets the Tonal be controlled by either a phone or tablet.

What’s the idea behind this project? What some may not know is that in addition serving home users, Tonal also partners with physical therapy clinics to help patients recover and rebuild. Clinics love Tonal for it’s ability to increment one pound at a time across hundreds of movements, in addition to the rich, multidimensional data about how users are moving.

This new project lets clinicians control the trainer from a phone or tablet: adjusting weight, switching modes, or pausing sessions on the fly. The goal is to improve workflow and make Tonal easier to use during hands-on treatment sessions. While the patient is working out, the physical therapist can monitor their performance and adjust as necessary from a dedicated control panel in the mobile app. The mobile app communicates with the Tonal via an API controlling a websocket connection. This could be expanded to the future to allow one mobile app to control multiple trainers for a group setting.

Mobile App Remote Control For Tonal

Tonal at Databricks Data + AI Summit

Data Science Team Presenting on Training Goals at the Databricks Summit

Tonal uses Databricks for some of our Data Science and reporting infrastructure, including calculating our users’ training goal progress and results. Giuseppe and Kristi were invited to speak about our architecture. See their full presentation here. Over 1,000 people experienced the Tonal Trainer firsthand over the three days. Our booth was buzzing with energy and excitement as fans lined up to try it for themselves.

We got to meet lots of Tonal fans at the conference! Thanks for the great vibes.

Fun

BBQ! Mini Golf! Skateboarding! From trying out a bunch of great restaurants in San Francisco to learning how to paint, it was a great time to connect with the Tonal team in person and have some fun.

“You stand on this part of the skateboard”

Is Tonal powered by Alien Technology from a recovered UFO? (No)

This entire course was actually very hard

Conclusion

The onsite was a success! Not just fun, but also productive, and it’s always great to get all of the teams together in one place. Congratulations again to our Hackathon winners.