From metrics to muscle: AI evolution in fitness tech

Executive Director & Partner
Executive Technology Director & Partner

Beyond steps and scores

The term “quantified self,” coined in 2007 by Wired editors Gary Wolf and Kevin Kelly, refers to using technology to track and analyze aspects of daily life, like steps, sleep, and heart rate. Early products like Fitbit paved the way, with advances in sensor tech and integration into smartphones and wearables accelerating the trend. Algorithms soon turned raw data into composite scores, such as Whoop’s Strain and Recovery Score, which track trends rather than specific units, helping users gauge effort and recovery over time.

Despite progress, we’re still grappling with moving from data to action. While quantified self tools offer self-knowledge and curiosity-driven insights, true behavior change hinges on knowing what to change and how to change it.

Smart Design’s work with Gatorade and Epicore on the Gx Sweat Patch shows how wearables are most powerful when they move beyond raw data. While the patch can track fluid loss and sodium levels, those numbers aren’t the point; most users wouldn’t remember them. What matters is what to do next. The real value comes from the patch’s personalized guidance: how much to drink and how to replenish electrolytes, both tailored to your workout and the weather. In this case, the data serves the recommendation, not the other way around.

The Gx Sweat Patch shows how products can deliver meaningful change by translating data into guidance, not just charts. With AI now enabling even deeper insights, we’re entering a new era of tech that doesn’t just track behavior—it transforms it.
Tucker Fort
Partner and Executive Director

The passive power of AI in fitness

With the abundance of new connected equipment integrating AI into fitness experiences, the market is evolving rapidly. When speaking directly with athletes and trainers, we were surprised to discover the ways they actually want to leverage these tools.

Rather than be used as a live coach or replace the in-the-moment support of a personal trainer, AI is most valuable when it stays in the background—enhancing workouts without distracting from them. Tasks like rep counting or auto-adjusting weights can keep athletes in the zone while still adding value.
 
A passive AI assistant allows users to stay focused on performance during the session and review rich data around reps, velocity, and weight afterward, when they’re in a mindset to analyze and improve. Compared to logging in a notebook, AI tracking offers deeper, more consistent insights over time. It can flag when it’s time to increase weight or make suggestions about number of reps. 
 
That said, the best way to get the most out of a workout is proper form and technique.  When paired with computer vision, a more active AI can step in like a virtual trainer, correcting form and stance in real-time to reduce injury risk and ensure the exercise is hitting the right muscle groups. This makes for a powerful opportunity for AI to complement human coaches, especially in group workout classes where the coach can’t be everywhere at once.
Whether passive or active, the value of AI in fitness hinges on accuracy. For recommendations to be trusted and useful, the data must be precise, ideally better than what athletes could track on their own.
Danielle Frucci
Executive Design Director

Competitive metrics vs personalized progress

There are countless ways to stay fit, some more intense than others, with strong communities built around formats like CrossFit, Peloton, and Pilates. CrossFit, for example, champions two key values: improving athleticism through varied, functional movement and doing it within a tight-knit community. And for the most part, it delivers.
 
But even within these communities, many start to question whether the goals—and the gym culture itself—are truly effective. Traditional markers like weight lifted or reps completed are easy to track, but they often push competitiveness over long-term progress. This can lead to injury, burnout, or performance plateaus. 
 
Similar to the rise of baseball analytics that challenged traditional scouting and helped move the sport to a data-driven, “Moneyball” approach, fitness is undergoing a similar transformation. We’re realizing that what we’ve measured in workouts doesn’t always reflect meaningful progress. Lifting the most weight or running the fastest mile isn’t consistently aligned with the purpose of the workout.
 
New technologies give us better ways to understand physical performance in real time. AI, wearable sensors, and adaptive resistance tools can now assess metrics like acceleration, heart rate, and muscle fatigue as you move, not just after the fact. This means workouts can be adjusted on the fly to match the intended goal, such as staying in a certain heart-rate zone or hitting the right speed during an anaerobic interval.
 
These aren’t necessarily “personalized” in the one-to-one sense, especially in class formats, but they are smarter. They respond to how you’re performing in the moment rather than relying on a fixed prescription. This shift is showing up in consumer wearables too. Devices like Garmin watches and Whoop bands are starting to incorporate anaerobic metrics once reserved for elite athletes, bringing more nuanced performance feedback to everyday users.
 
We’ve already seen how this real-time adaptability benefits professional sports. In our Gx Sweat Patch research, we worked with elite soccer teams who tracked total distance run using GPS sensors in their cleats. When a player exceeded training thresholds, coaches would manually adjust workloads. Now, AI can automate these changes based on a broader range of inputs, optimizing training while reducing injury risk.
It’s time we reframe what progress looks like in fitness. Instead of focusing on static numbers, the feedback loop can become feel, adapt, and improve. With technology, the goal isn’t just to do more—it’s to do what matters, better.
John Anderson
Partner and Executive Technology Director

Let’s design a smarter world together