The AI Personal Trainer Is Coming

Open AI launched its first large-scale brand campaign in September. Three commercials each highlight a different ChatGPT use case: cooking, travel and fitness.

At Will Ventures, we got particularly excited about the fitness-themed ad, which features a young man asking ChatGPT for a workout plan. Why did we get excited? Because we strongly believe a new AI fitness coaching app will break out in the next few years, and we don’t think it will be ChatGPT.

Personal trainers are a $46B global industry – but they’re a luxury service. The average personal trainer in the US charges ~$60/hour, and the best trainers charge hundreds if not thousands per hour. Consumers who can afford it love it. But the TAM’s constrained by the price point.

In many ways, personal fitness is an ideal consumer use case for AI. Every consumer has different health goals, injuries, time pressures and access to equipment – and AI can personalize workout plans around those needs. Health data is increasingly abundant, as wearables grow more prevalent and diagnostics testing costs come down – and AI can synthesize that data.

AI can also create the type of continuous action-to-feedback loops that consumer operators and investors love. The consumer receives a personalized workout plan, performs the workout as it’s logged, and receives continuous feedback from their wearables (and perhaps computer vision tracking), which informs their future personalized workouts.

AI can now do most things a personal trainer does (and do some of them even better), all at a cost that’s accessible to the masses.

Open AI, Google and other large foundational model owners won’t be the companies to build this product, for a few reasons. A dedicated fitness app is more functional than a ChatGPT “everything app” where you have to sift through your cooking and travel queries. Also, branding is paramount in fitness. Consumers don’t just want to work out; they want to feel emotionally rewarded for working out – and aligned with fitness brands that speak to them. It’d be a distraction for Open AI or Google to spin up a separate fitness brand. Lastly, the scrappiest startups will win out over the giants, considering building a fitness app doesn’t require significant access to capital. It requires listening to users and iterating on product quickly.

Just last week Open AI announced ChatGPT Health, saying: “You can now securely connect medical records and wellness apps—like Apple Health, Function, and MyFitnessPal—so ChatGPT can help you understand recent test results, prepare for appointments with your doctor, get advice on how to approach your diet and workout routine, or understand the tradeoffs of different insurance options based on your healthcare patterns.” In other words, Open AI is looking to aggregate all these health-related apps, not build them themselves.

So, who will build the dominant AI-powered personal trainer app?

We think it’s helpful to look at the current market of fitness apps, which break down into roughly three categories.

1. Content library apps like Peloton and FitOn offer libraries of workout videos. You might follow a fitness influencer in a studio for a 30-minute workout. These apps win based on sleek branding and product, and you have plenty of workouts to choose from. But the workouts aren’t tailored to your specific fitness needs.

2. Exercise logging apps like Fitbod, Strong and GymVerse let you log workouts, mostly for weight training. These products are popular but don’t have a huge moat, so you’ll find a ton of exercise logging apps on the App Store.

3. Activity-based apps like Strava, AllTrails and Slopes track exercise for one specific activity. These companies have come to dominate their respective activities, largely thanks to their social features and engagement loops. But these apps don’t serve as all-in-one fitness companions.

Several apps have broken out recently by combining elements across these three categories. Runna is an AI-driven run-coaching app that personalizes training plans by pace, goals, and race calendar. They reached hundreds of thousands of users before being acquired by Strava. Ladder is a strength training app where users choose their ‘coach’, and are led through cohesive programs week-by-week with content tailored to that coach’s specialty. They’re currently the second top-grossing health & fitness app on the US App Store. Both of these apps pair great software with strong branding and social features.

We also think it’s helpful to understand that there will be real challenges in building a breakout AI-powered personal trainer app. For simplicity’s sake, we’ll focus on the three most glaring challenges, as we see them.

Challenge #1 - Product Positioning

Consumer fitness is crowded and spend is fragmented across gym memberships, online / offline coaching, at-home hardware, video content, and wearables. It remains unclear which of these categories AI training apps may be able to fully replace vs compliment. When consumers are bogged down with choices, precise product positioning is critical to drive trial and justify another subscription.

Challenge #2 - Willingness to Pay

Building off of #1, the scale required to achieve venture-type outcomes at a $10/month price point is monumental. There will be numerous mass market offerings competing away margin, and LTVs are notoriously difficult in digital fitness due to the flakiness of casual athletes. If you pursue the other side of the curve, fitness nuts are more comfortable spending but have a high bar for product utility + outcomes. Can you actually deliver a “premium” software experience that justifies a $30-50/month price point?

Challenge #3 - Retention

Removing the human-in-the-loop eliminates one of the great forcing functions in fitness. Will anyone be afraid to skip their workout and have to justify it to their AI coach? Product mechanisms aimed at ongoing engagement, community, and accountability won’t be easy to layer into an AI coaching experience, but they’ll be critical for sustained growth.

How does an app overcome these challenges? In our opinion, there’s no silver bullet.

The winning app will need to combine the brand power of a Peloton, the functionality of a Fitbod, and the social & engagement loops of a Strava – and all while adding an AI personalization layer that mirrors a personal trainer. The winning team will be one that’s fast to iterate on product and obsessed with listening to users.

We’re constantly trying out new fitness apps that launch on the App Store. If you’re building an AI-powered fitness app, please reach out! We’d love to try out your product.