Camille Duncan, left, and Michelle Siman use app- and AI-based tech to train and meet fitness goals.Tobias Wang/The Globe and Mail
In 2018, Matt Spettel saw few wearable tech products geared to strength training. So he and his childhood friend Gabe Madonna developed their own, using AI to measure motion and activity and give us feedback on their pacing and form.
The initial response was tepid. “We learned that just tracking what people were doing in the gym was cool and tech-y, but it wasn’t really solving the core problem,” Spettel says. “They didn’t have a strong sense of what to do and weren’t experts in strength training.”
In other words, AI wasn’t enough to keep amateur athletes motivated to use the app.
At the end of 2020, Spettel and Madonna launched Trainwell (then called CoPilot), a platform that pairs people with a remote personal trainer who personalizes plans to help users achieve their fitness goals. Engagement spiked tenfold. “We found that the best way to make plans and get people to do that was with human intervention,” says Spettel. Today, Trainwell uses AI to strengthen the relationship between user and coach. For example, trainers use AI software to transcribe and synthesize client information to better personalize training plans; Trainwell uses a different AI model to emulate trainers’ voices during workouts so that it sounds like they’re in the room giving immediate feedback and guidance. Spettel pegs the company’s annual revenue in the eight figures.
Trainwell’s success proves what sports psychology experts suspect: On its own, AI can’t replace human interaction. But it can be a powerful tool to help build community, offer accountability and motivate fitness enthusiasts. The popular running app Strava and Whoop, wearable tracking technology, use data-driven insights to offer feedback, personalize workouts and build community with other athletes. These tools don’t just monitor performance and offer tailored advice but aim to add an element of professional coaching to the average person’s fitness routine.
Serious amateur athletes use both wearable tech and training apps to achieve their fitness goals or enhance their performance. Camille Duncan, an art director in Toronto, uses a Garmin Forerunner 55 watch to measure her heart rate, calories lost, length and pace during her high-intensity interval training (HIIT) and runs. She also uses Garmin Coach, a free feature where one of the company’s three coaches will personalize a training plan to help users improve their fitness or reach a goal. The workouts are synced to the watch and adapt based on performance. “I use it as a way to hold accountability for my training plans,” says Duncan of her assigned coach, Greg McMillan, who created a plan that included speed workouts before her recent half-marathon.
Rather than deliver generic feedback, good AI should build algorithms that use data to support behaviour change “to work with users’ motivation and change their view of self,” says Dr. Nicole Culos-Reed, a kinesiology professor at the University of Calgary who studies health and exercise psychology.
The fitness platform Trainwell pairs amateur athletes and fitness enthusiasts with trainers who create plans and coach them remotely.Supplied
Building parameters into AI-powered apps is another challenge to ensure people are training safely. “If it’s only an AI model, it’s hard to integrate feedback from the user,” says Dr. Daniel Fuller, an associate professor in the Department of Community Health and Epidemiology at the University of Saskatchewan who researches wearable fitness technologies. “There’s pretty high potential for over-training, if you don’t know how to adapt your body to the training.”
That’s where recovery metrics matter. Michelle Siman, a marketing director in Toronto, appreciates that Whoop isn’t “just trying to get you addicted to hitting a certain number, like calories or steps,” like other trackers. Whoop, she says, “allows you to focus on recovery every day to perform optimally,” by tracking wellness metrics including skin temperature, sleep performance and resting heart. Last year, Whoop introduced an AI-powered coaching feature that uses OpenAI to generate tailored nutrition advice and fitness plans based on user data.
Even apps with coaching plans should adapt their training plans to athletes’ needs to avoid users experiencing an injury or quitting altogether. To train for a triathlon, for example, Fuller signed up for a premium plan on TriDot, Ironman’s training app, which offers AI-powered training paired with a dedicated coach. Fuller’s multisport GPS watch, the Garmin Fenix 5, would share his biometric and performance data to TriDot app, while a coach would check in and adapt his plan as needed. Still, Fuller found his swim program too difficult to continue. “The swims were always high intensity and there was almost no easy swimming,” he says. “So the swims became very taxing and I struggled to manage the other workouts on the program.”
Dr. Culos-Reed believes that rather than simply optimizing physical performance, trainers – real or virtual – should understand athletes’ attitudes and mindsets about exercising. Apps need to tap into “the emotion side, the affect” to exercise behaviour, she says. “If you look at sport, a lot of coaching is from the negative. It’s ‘You’re doing this wrong.’” Apps that offer positive reinforcement and support may just be key to keep a first-time weightlifter going. “I’m talking not just getting somebody started but keeping somebody going,” she says. “Maintenance.”
